diff --git a/docs/configuration.rst b/docs/configuration.rst
index 28607c22..7d73f508 100644
--- a/docs/configuration.rst
+++ b/docs/configuration.rst
@@ -2,57 +2,206 @@ Configuration Reference
 =======================
 
 ``towncrier`` has many knobs and switches you can use, to customize it to your project's needs.
-The setup in the `Tutorial <tutorial.html>`_ doesn't touch on many, but this document will detail each of these options for you!
+The setup in the :doc:`tutorial` doesn't touch on many, but this document will detail each of these options for you!
 
-For how to perform common customization tasks, see `Customization <customization/index.html>`_.
+For how to perform common customization tasks, see :doc:`customization/index`.
 
 ``[tool.towncrier]``
 --------------------
 
-All configuration for ``towncrier`` sits inside ``pyproject.toml``, under the ``tool.towncrier`` namespace.
+All configuration for ``towncrier`` sits inside ``towncrier.toml`` or ``pyproject.toml``, under the ``tool.towncrier`` namespace.
 Please see https://toml.io/ for how to write TOML.
 
+A minimal configuration for a Python project looks like this:
+
+.. code-block:: toml
+
+   # pyproject.toml
+
+   [tool.towncrier]
+   package = "myproject"
+
+A minimal configuration for a non-Python project looks like this:
+
+.. code-block:: toml
+
+   # towncrier.toml
+
+   [tool.towncrier]
+   name = "My Project"
 
 Top level keys
 ~~~~~~~~~~~~~~
 
-- ``directory`` -- If you are not storing your news fragments in your Python package, or aren't using Python, this is the path to where your newsfragments will be put.
-- ``filename`` -- The filename of your news file.
-  ``NEWS.rst`` by default.
-- ``package`` -- The package name of your project.
-  (Python projects only)
-- ``package_dir`` -- The folder your package lives. ``./`` by default, some projects might need to use ``src``.
-  (Python projects only)
-- ``template`` -- Path to an alternate template for generating the news file, if you have one.
-- ``start_string`` -- The magic string that ``towncrier`` looks for when considering where the release notes should start.
-  ``.. towncrier release notes start`` by default.
-- ``title_format`` -- A format string for the title of your project.
-  ``{name} {version} ({project_date})`` by default.
-- ``issue_format`` -- A format string for rendering the issue/ticket number in newsfiles.
-  ``#{issue}`` by default.
-- ``underlines`` -- The characters used for underlining headers.
-  ``["=", "-", "~"]`` by default.
+``name``
+    The name of your project.
+
+    For Python projects that provide a ``package`` key, if left empty then the name will be automatically determined.
+
+    ``""`` by default.
+
+``version``
+    The version of your project.
+
+    Python projects that provide the ``package`` key can have the version to be automatically determined from a ``__version__`` variable in the package's module.
+
+    If not provided or able to be determined, the version must be passed explicitly by the command line argument ``--version``.
+
+``directory``
+    The directory storing your news fragments.
+
+    For Python projects that provide a ``package`` key, the default is a ``newsfragments`` directory within the package.
+    Otherwise the default is a ``newsfragments`` directory relative to the configuration file.
+
+``filename``
+    The filename of your news file.
+
+    ``"NEWS.rst"`` by default.
+
+``template``
+    Path to the template for generating the news file.
+
+    If the path looks like ``<some.package>:<filename.ext>``, it is interpreted as a template bundled with an installed Python package.
+
+    ``"towncrier:default.rst"`` by default unless ``filename`` ends with ``.md``, in which case the default is ``"towncrier:default.md"``.
+
+``start_string``
+    The magic string that ``towncrier`` looks for when considering where the release notes should start.
+
+    ``".. towncrier release notes start\n"`` by default unless ``filename`` ends with ``.md``, in which case the default is ``"<!-- towncrier release notes start -->\n"``.
+
+``title_format``
+    A format string for the title of your project.
+
+    The explicit value of ``False`` will disable the title entirely.
+    Any other empty value means the template should render the title (the bundled templates use ``<name> <version> (<date>)``).
+    Strings should use the following keys to render the title dynamically: ``{name}``, ``{version}``, and ``{project_date}``.
+
+    ``""`` by default.
+
+``issue_format``
+    A format string for rendering the issue/ticket number in newsfiles.
+
+    If none, the issues are rendered as ``#<issue>`` if for issues that are integers, or just ``<issue>`` otherwise.
+    Use the ``{issue}`` key in your string render the issue number, for example Markdown projects may want to use ``"[{issue}]: https://<your bug tracker>/{issue}"``.
+
+    ``None`` by default.
+
+``underlines``
+    The characters used for underlining headers.
+
+    Not used in the bundled Markdown template.
+
+    ``["=", "-", "~"]`` by default.
+
+``wrap``
+    Boolean value indicating whether to wrap news fragments to a line length of 79.
+
+    ``false`` by default.
+
+``all_bullets``
+    Boolean value indicating whether the template uses bullets for each news fragment.
+
+    ``true`` by default.
+
+``single_file``
+    Boolean value indicating whether to write all news fragments to a single file.
+
+    If ``false``, the ``filename`` should use the following keys to render the filenames dynamically:
+    ``{name}``, ``{version}``, and ``{project_date}``.
+
+    ``true`` by default.
+
+``orphan_prefix``
+    The prefix used for orphaned news fragments.
+
+    ``"+"`` by default.
+
+Extra top level keys for Python projects
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+``package``
+    The Python package name of your project.
+
+    Allows ``name`` and ``version`` to be automatically determined from the Python package.
+    Changes the default ``directory`` to be a ``newsfragments`` directory within this package.
+
+``package_dir``
+    The folder your package lives.
+
+    ``"."`` by default, some projects might need to use ``"src"``.
+
+
+Sections
+--------
+
+``towncrier`` supports splitting fragments into multiple sections, each with its own news of fragment types.
+
+Add an array of tables your ``.toml`` configuration file named ``[[tool.towncrier.section]]``.
+
+Each table within this array has the following mandatory keys:
+
+
+``name``
+    The name of the section.
+
+``path``
+    The path to the directory containing the news fragments for this section, relative to the configured ``directory``.
+    Use ``""`` for the root directory.
+
+For example:
+
+.. code-block:: toml
+
+   [[tool.towncrier.section]]
+   name = "Main Platform"
+   path = ""
+
+   [[tool.towncrier.section]]
+   name = "Secondary"
+   path = "secondary"
+
+Section Path Behaviour
+~~~~~~~~~~~~~~~~~~~~~~
+
+The path behaviour is slightly different depending on whether ``directory`` is explicitly set.
+
+If ``directory`` is not set, "newsfragments" is added to the end of each path. For example, with the above sections, the paths would be:
+
+:Main Platform:  ./newsfragments
+:Secondary:      ./secondary/newsfragments
+
+If ``directory`` *is* set, the section paths are appended to this path. For example, with ``directory = "changes"`` and the above sections, the paths would be:
+
+:Main Platform:  ./changes
+:Secondary:      ./changes/secondary
 
 
 Custom fragment types
 ---------------------
-``towncrier`` allows defining custom fragment types.
-Custom fragment types will be used instead ``towncrier`` default ones, they are not combined.
 
-There are two ways to add custom fragment types.
+``towncrier`` has the following default fragment types: ``feature``, ``bugfix``, ``doc``, ``removal``, and ``misc``.
+
+You can use either of the two following method to define custom types instead (you will need to redefine any of the default types you want to use).
 
 
-Defining Custom Fragment Types With a TOML Mapping
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Use TOML tables (alphabetical order)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-Users can configure each of their own custom fragment types by adding tables to
-the pyproject.toml named ``[tool.towncrier.fragment.<a custom fragment type>]``.
+Adding tables to your ``.toml`` configuration file named ``[tool.towncrier.fragment.<a custom fragment type>]``.
 
-These tables may include the following optional keys:
+These may include the following optional keys:
 
- * ``name``: The description of the fragment type, as it must be included in the news file.
-   If omitted, it defaults to  its  fragment type, but capitalized.
- * ``showcontent``: Whether if the fragment contents should be included in the news file. If omitted, it defaults to ``true``
+
+``name``
+    The description of the fragment type, as it must be included in the news file.
+
+    Defaults to its fragment type, but capitalized.
+
+``showcontent``
+    A boolean value indicating whether the fragment contents should be included in the news file.
+
+    ``true`` by default.
 
 For example, if you want your custom fragment types to be ``["feat", "fix", "chore",]`` and you want all of them to use the default configuration except ``"chore"`` you can do it as follows:
 
@@ -70,25 +219,30 @@ For example, if you want your custom fragment types to be ``["feat", "fix", "cho
 
 .. warning::
 
-   Since TOML mappings aren't ordered, the sections are always rendered alphabetically.
+   Since TOML mappings aren't ordered, types defined using this method are always rendered alphabetically.
+
 
+Use a TOML Array (defined order)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-Defining Custom Fragment Types With an Array of TOML Tables
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Add an array of tables to your ``.toml`` configuration file named ``[[tool.towncrier.type]]``.
 
-Users can create their own custom fragment types by adding an array of
-tables to the pyproject.toml named ``[[tool.towncrier.type]]``.
+If you use this way to configure custom fragment types, ensure there is no ``tool.towncrier.fragment`` table.
 
-If you use this way to configure custom fragment types, please note that ``fragment_types`` must be empty or not provided.
+Each table within this array has the following mandatory keys:
 
-Each custom type (``[[tool.towncrier.type]]``) has the following
-mandatory keys:
 
-* ``directory``: The type / category of the fragment.
-* ``name``: The description of the fragment type, as it must be included
-  in the news file.
-* ``showcontent``: Whether if the fragment contents should be included in the
-  news file.
+``directory``
+    The type / category of the fragment.
+
+``name``
+    The description of the fragment type, as it must be included
+    in the news file.
+
+``showcontent``
+    A boolean value indicating whether the fragment contents should be included in the news file.
+
+    ``true`` by default.
 
 For example:
 
@@ -104,40 +258,3 @@ For example:
    directory = "chore"
    name = "Other Tasks"
    showcontent = false
-
-
-All Options
------------
-
-``towncrier`` has the following global options, which can be specified in the toml file:
-
-.. code-block:: toml
-
-   [tool.towncrier]
-   package = ""
-   package_dir = "."
-   single_file = true  # if false, filename is formatted like `title_format`.
-   filename = "NEWS.rst"
-   directory = "directory/of/news/fragments"
-   version = "1.2.3"  # project version if maintained separately
-   name = "arbitrary project name"
-   template = "path/to/template.rst"
-   start_string = "Text used to detect where to add the generated content in the middle of a file. Generated content added after this text. Newline auto added."
-   title_format = "{name} {version} ({project_date})"  # or false if template includes title
-   issue_format = "format string for {issue} (issue is the first part of fragment name)"
-   underlines = "=-~"
-   wrap = false  # Wrap text to 79 characters
-   all_bullets = true  # make all fragments bullet points
-   orphan_prefix = "+"   # Prefix for orphan news fragment files, set to "" to disable.
-
-If ``single_file`` is set to ``true`` or unspecified, all changes will be written to a single fixed newsfile, whose name is literally fixed as the ``filename`` option.
-In each run of ``towncrier build``, content of new changes will append at the top of old content, and after ``start_string`` if the ``start_string`` already appears in the newsfile.
-If the corresponding ``top_line``, which is formatted as the option 'title_format', already exists in newsfile, ``ValueError`` will be raised to remind you "already produced newsfiles for this version".
-
-If ``single_file`` is set to ``false`` instead, each versioned ``towncrier build`` will generate a separate newsfile, whose name is formatted as the pattern given by option ``filename``.
-For example, if ``filename="{version}-notes.rst"``, then the release note with version "7.8.9" will be written to the file "7.8.9-notes.rst".
-If the newsfile already exists, its content will be overwritten with new release note, without throwing a ``ValueError`` warning.
-
-If ``title_format`` is unspecified or an empty string, the default format will be used.
-If set to ``false``, no title will be created.
-This can be useful if the specified template creates the title itself.
diff --git a/docs/customization/newsfile.rst b/docs/customization/newsfile.rst
index dc5a7935..67f3df4f 100644
--- a/docs/customization/newsfile.rst
+++ b/docs/customization/newsfile.rst
@@ -5,23 +5,36 @@ Adding Content Above ``towncrier``
 ----------------------------------
 
 If you wish to have content at the top of the news file (for example, to say where you can find the tickets), you can use a special rST comment to tell ``towncrier`` to only update after it.
-In your existing news file (e.g. ``NEWS.rst``), add the following line above where you want ``towncrier`` to put content::
+In your existing news file (e.g. ``NEWS.rst``), add the following line above where you want ``towncrier`` to put content:
 
-  .. towncrier release notes start
+.. code-block:: restructuredtext
 
-In an existing news file, it'll look something like this::
+    .. towncrier release notes start
 
-  This is the changelog of my project. You can find the
-  issue tracker at http://blah.
+In an existing news file, it'll look something like this:
 
-  .. towncrier release notes start
+.. code-block:: restructuredtext
 
-  myproject 1.0.2 (2018-01-01)
-  ============================
+    This is the changelog of my project. You can find the
+    issue tracker at http://blah.
 
-  Bugfixes
-  --------
+    .. towncrier release notes start
 
-  - Fixed, etc...
+    myproject 1.0.2 (2018-01-01)
+    ============================
+
+    Bugfixes
+    --------
+
+    - Fixed, etc...
 
 ``towncrier`` will not alter content above the comment.
+
+Markdown
+~~~~~~~~
+
+If your news file is in Markdown (e.g. ``NEWS.md``), use the following comment instead:
+
+.. code-block:: html
+
+    <!-- towncrier release notes start -->
diff --git a/docs/tutorial.rst b/docs/tutorial.rst
index 7b893b1d..0a42ff35 100644
--- a/docs/tutorial.rst
+++ b/docs/tutorial.rst
@@ -1,7 +1,7 @@
 Tutorial
 ========
 
-This tutorial assumes you have a Python project with a *reStructuredText* (rst) news file (also known as changelog) file that you wish to use ``towncrier`` on, to generate its news file.
+This tutorial assumes you have a Python project with a *reStructuredText* (rst) or *Markdown* (md) news file (also known as changelog) that you wish to use ``towncrier`` on, to generate its news file.
 It will cover setting up your project with a basic configuration, which you can then feel free to `customize <customization/index.html>`_.
 
 Install from PyPI::
@@ -144,6 +144,8 @@ You should get an output similar to this::
 
    - #1, #2
 
+Note: if you configure a Markdown file (for example, ``filename = "CHANGES.md"``) in your configuration file, the titles will be output in Markdown format instead.
+
 
 Producing News Files In Production
 ----------------------------------
@@ -161,6 +163,8 @@ If you wish to have content at the top of the news file (for example, to say whe
 
 ``towncrier`` will then put the version notes after this comment, and leave your existing content that was above it where it is.
 
+Note: if you configure a Markdown file (for example, ``filename = "CHANGES.md"``) in your configuration file, the comment should be ``<!-- towncrier release notes start -->`` instead.
+
 
 Finale
 ------
diff --git a/pyproject.toml b/pyproject.toml
index d54647d1..8738e59b 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -32,7 +32,7 @@ requires-python = ">=3.7"
 dependencies = [
     "click",
     "click-default-group",
-    "importlib-resources>1.3; python_version<'3.9'",
+    "importlib-resources>=5; python_version<'3.10'",
     "incremental",
     "jinja2",
     "tomli; python_version<'3.11'",
@@ -136,6 +136,12 @@ profile = "attrs"
 line_length = 88
 
 
+[tool.ruff.isort]
+# Match isort's "attrs" profile
+lines-after-imports = 2
+lines-between-types = 1
+
+
 [tool.mypy]
 strict = true
 # 2022-09-04: Trial's API isn't annotated yet, which limits the usefulness of type-checking
diff --git a/src/towncrier/_builder.py b/src/towncrier/_builder.py
index f6cd87bd..d1b14dba 100644
--- a/src/towncrier/_builder.py
+++ b/src/towncrier/_builder.py
@@ -8,7 +8,7 @@
 import textwrap
 import traceback
 
-from collections import OrderedDict, defaultdict
+from collections import defaultdict
 from typing import Any, DefaultDict, Iterable, Iterator, Mapping, Sequence
 
 from jinja2 import Template
@@ -66,15 +66,14 @@ def parse_newfragment_basename(
 
 # Returns a structure like:
 #
-# OrderedDict([
-#   ("",
-#    {
-#      ("142", "misc"): u"",
-#      ("1", "feature"): u"some cool description",
-#    }),
-#   ("Names", {}),
-#   ("Web", {("3", "bugfix"): u"Fixed a thing"}),
-# ])
+# {
+#     "": {
+#         ("142", "misc"): "",
+#         ("1", "feature"): "some cool description",
+#     },
+#     "Names": {},
+#     "Web": {("3", "bugfix"): "Fixed a thing"},
+# }
 #
 # We should really use attrs.
 #
@@ -89,7 +88,7 @@ def find_fragments(
     """
     Sections are a dictonary of section names to paths.
     """
-    content = OrderedDict()
+    content = {}
     fragment_filenames = []
     # Multiple orphan news fragments are allowed per section, so initialize a counter
     # that can be incremented automatically.
@@ -164,7 +163,7 @@ def split_fragments(
     definitions: Mapping[str, Mapping[str, Any]],
     all_bullets: bool = True,
 ) -> Mapping[str, Mapping[str, Mapping[str, Sequence[str]]]]:
-    output = OrderedDict()
+    output = {}
 
     for section_name, section_fragments in fragments.items():
         section: dict[str, dict[str, list[str]]] = {}
@@ -182,7 +181,7 @@ def split_fragments(
             if definitions[category]["showcontent"] is False:
                 content = ""
 
-            texts = section.setdefault(category, OrderedDict())
+            texts = section.setdefault(category, {})
 
             tickets = texts.setdefault(content, [])
             if ticket:
@@ -255,12 +254,15 @@ def render_fragments(
 
     jinja_template = Template(template, trim_blocks=True)
 
-    data: dict[str, dict[str, dict[str, list[str]]]] = OrderedDict()
+    data: dict[str, dict[str, dict[str, list[str]]]] = {}
+    issues_by_category: dict[str, dict[str, list[str]]] = {}
 
     for section_name, section_value in fragments.items():
-        data[section_name] = OrderedDict()
+        data[section_name] = {}
+        issues_by_category[section_name] = {}
 
         for category_name, category_value in section_value.items():
+            category_issues: set[str] = set()
             # Suppose we start with an ordering like this:
             #
             # - Fix the thing (#7, #123, #2)
@@ -273,6 +275,7 @@ def render_fragments(
             entries = []
             for text, issues in category_value.items():
                 entries.append((text, sorted(issues, key=issue_key)))
+                category_issues.update(issues)
 
             # Then we sort the lines:
             #
@@ -284,12 +287,16 @@ def render_fragments(
 
             # Then we put these nicely sorted entries back in an ordered dict
             # for the template, after formatting each issue number
-            categories = OrderedDict()
+            categories = {}
             for text, issues in entries:
                 rendered = [render_issue(issue_format, i) for i in issues]
                 categories[text] = rendered
 
             data[section_name][category_name] = categories
+            issues_by_category[section_name][category_name] = [
+                render_issue(issue_format, i)
+                for i in sorted(category_issues, key=issue_key)
+            ]
 
     done = []
 
@@ -311,6 +318,7 @@ def get_indent(text: str) -> str:
         versiondata=versiondata,
         top_underline=top_underline,
         get_indent=get_indent,  # simplify indentation in the jinja template.
+        issues_by_category=issues_by_category,
     )
 
     for line in res.split("\n"):
diff --git a/src/towncrier/_settings/fragment_types.py b/src/towncrier/_settings/fragment_types.py
index d370986d..a9882c9a 100644
--- a/src/towncrier/_settings/fragment_types.py
+++ b/src/towncrier/_settings/fragment_types.py
@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 import abc
-import collections as clt
 
 from typing import Any, Iterable, Mapping
 
@@ -36,16 +35,14 @@ def load(self) -> Mapping[str, Mapping[str, Any]]:
 class DefaultFragmentTypesLoader(BaseFragmentTypesLoader):
     """Default towncrier's fragment types."""
 
-    _default_types = clt.OrderedDict(
-        [
-            # Keep in-sync with docs/tutorial.rst.
-            ("feature", {"name": "Features", "showcontent": True}),
-            ("bugfix", {"name": "Bugfixes", "showcontent": True}),
-            ("doc", {"name": "Improved Documentation", "showcontent": True}),
-            ("removal", {"name": "Deprecations and Removals", "showcontent": True}),
-            ("misc", {"name": "Misc", "showcontent": False}),
-        ]
-    )
+    _default_types = {
+        # Keep in-sync with docs/tutorial.rst.
+        "feature": {"name": "Features", "showcontent": True},
+        "bugfix": {"name": "Bugfixes", "showcontent": True},
+        "doc": {"name": "Improved Documentation", "showcontent": True},
+        "removal": {"name": "Deprecations and Removals", "showcontent": True},
+        "misc": {"name": "Misc", "showcontent": False},
+    }
 
     def load(self) -> Mapping[str, Mapping[str, Any]]:
         """Load default types."""
@@ -72,7 +69,7 @@ class ArrayFragmentTypesLoader(BaseFragmentTypesLoader):
     def load(self) -> Mapping[str, Mapping[str, Any]]:
         """Load types from toml array of mappings."""
 
-        types = clt.OrderedDict()
+        types = {}
         types_config = self.config["type"]
         for type_config in types_config:
             directory = type_config["directory"]
@@ -123,7 +120,7 @@ def load(self) -> Mapping[str, Mapping[str, Any]]:
             (fragment_type, self._load_options(fragment_type))
             for fragment_type in fragment_types
         ]
-        types = clt.OrderedDict(custom_types_sequence)
+        types = dict(custom_types_sequence)
         return types
 
     def _load_options(self, fragment_type: str) -> Mapping[str, Any]:
diff --git a/src/towncrier/_settings/load.py b/src/towncrier/_settings/load.py
index 10bde248..92098f9d 100644
--- a/src/towncrier/_settings/load.py
+++ b/src/towncrier/_settings/load.py
@@ -4,13 +4,14 @@
 from __future__ import annotations
 
 import atexit
+import dataclasses
 import os
+import re
 import sys
 
-from collections import OrderedDict
 from contextlib import ExitStack
-from dataclasses import dataclass
-from typing import TYPE_CHECKING, Any, Mapping
+from pathlib import Path
+from typing import TYPE_CHECKING, Any, Mapping, Sequence
 
 from .._settings import fragment_types as ft
 
@@ -23,7 +24,7 @@
     else:
         from typing import Literal
 
-if sys.version_info < (3, 9):
+if sys.version_info < (3, 10):
     import importlib_resources as resources
 else:
     from importlib import resources
@@ -35,25 +36,28 @@
     import tomllib
 
 
-@dataclass
+re_resource_template = re.compile(r"[-\w.]+:[-\w.]+$")
+
+
+@dataclasses.dataclass
 class Config:
-    package: str
-    package_dir: str
-    single_file: bool
-    filename: str
-    directory: str | None
-    version: str | None
-    name: str | None
     sections: Mapping[str, str]
     types: Mapping[str, Mapping[str, Any]]
-    template: str
+    template: str | tuple[str, str]
     start_string: str
-    title_format: str | Literal[False]
-    issue_format: str | None
-    underlines: list[str]
-    wrap: bool
-    all_bullets: bool
-    orphan_prefix: str
+    package: str = ""
+    package_dir: str = "."
+    single_file: bool = True
+    filename: str = "NEWS.rst"
+    directory: str | None = None
+    version: str | None = None
+    name: str = ""
+    title_format: str | Literal[False] = ""
+    issue_format: str | None = None
+    underlines: Sequence[str] = ("=", "-", "~")
+    wrap: bool = False
+    all_bullets: bool = True
+    orphan_prefix: str = "+"
 
 
 class ConfigError(Exception):
@@ -62,12 +66,6 @@ def __init__(self, *args: str, **kwargs: str):
         super().__init__(*args)
 
 
-_start_string = ".. towncrier release notes start\n"
-_title_format = None
-_template_fname = "towncrier:default"
-_underlines = ["=", "-", "~"]
-
-
 def load_config_from_options(
     directory: str | None, config_path: str | None
 ) -> tuple[str, Config]:
@@ -118,82 +116,83 @@ def load_config_from_file(directory: str, config_file: str) -> Config:
 
 
 def parse_toml(base_path: str, config: Mapping[str, Any]) -> Config:
-    if "tool" not in config:
+    if "towncrier" not in (config.get("tool") or {}):
         raise ConfigError("No [tool.towncrier] section.", failing_option="all")
 
     config = config["tool"]["towncrier"]
+    parsed_data = {}
+
+    # Check for misspelt options.
+    for typo, correct in [
+        ("singlefile", "single_file"),
+    ]:
+        if config.get(typo):
+            raise ConfigError(
+                f"`{typo}` is not a valid option. Did you mean `{correct}`?",
+                failing_option=typo,
+            )
 
-    sections = OrderedDict()
+    # Process options.
+    for field in dataclasses.fields(Config):
+        if field.name in ("sections", "types", "template"):
+            # Skip these options, they are processed later.
+            continue
+        if field.name in config:
+            #  Interestingly, the __future__ annotation turns the type into a string.
+            if field.type in ("bool", bool):
+                if not isinstance(config[field.name], bool):
+                    raise ConfigError(
+                        f"`{field.name}` option must be boolean: false or true.",
+                        failing_option=field.name,
+                    )
+            parsed_data[field.name] = config[field.name]
+
+    # Process 'section'.
+    sections = {}
     if "section" in config:
         for x in config["section"]:
             sections[x.get("name", "")] = x["path"]
     else:
         sections[""] = ""
+    parsed_data["sections"] = sections
+
+    # Process 'types'.
     fragment_types_loader = ft.BaseFragmentTypesLoader.factory(config)
-    types = fragment_types_loader.load()
-
-    wrap = config.get("wrap", False)
-
-    single_file_wrong = config.get("singlefile")
-    if single_file_wrong:
-        raise ConfigError(
-            "`singlefile` is not a valid option. Did you mean `single_file`?",
-            failing_option="singlefile",
-        )
-
-    single_file = config.get("single_file", True)
-    if not isinstance(single_file, bool):
-        raise ConfigError(
-            "`single_file` option must be a boolean: false or true.",
-            failing_option="single_file",
-        )
-
-    all_bullets = config.get("all_bullets", True)
-    if not isinstance(all_bullets, bool):
-        raise ConfigError(
-            "`all_bullets` option must be boolean: false or true.",
-            failing_option="all_bullets",
-        )
-
-    template = config.get("template", _template_fname)
-    if template.startswith("towncrier:"):
-        resource_name = f"templates/{template.split('towncrier:', 1)[1]}.rst"
-        resource_path = _file_manager.enter_context(
-            resources.as_file(resources.files("towncrier") / resource_name)
-        )
-
-        if not resource_path.is_file():
+    parsed_data["types"] = fragment_types_loader.load()
+
+    # Process 'template'.
+    markdown_file = Path(config.get("filename", "")).suffix == ".md"
+    template = config.get("template", "towncrier:default")
+    if re_resource_template.match(template):
+        package, resource = template.split(":", 1)
+        if not Path(resource).suffix:
+            resource += ".md" if markdown_file else ".rst"
+        if not resources.is_resource(package, resource):
+            if resources.is_resource(package + ".templates", resource):
+                package += ".templates"
+            else:
+                raise ConfigError(
+                    f"'{package}' does not have a template named '{resource}'.",
+                    failing_option="template",
+                )
+        template = (package, resource)
+    else:
+        template = os.path.join(base_path, template)
+        if not os.path.isfile(template):
             raise ConfigError(
-                "Towncrier does not have a template named '%s'."
-                % (template.split("towncrier:", 1)[1],)
+                f"The template file '{template}' does not exist.",
+                failing_option="template",
             )
 
-        template = str(resource_path)
-    else:
-        template = os.path.join(base_path, template)
+    parsed_data["template"] = template
+
+    # Process 'start_string'.
+
+    start_string = config.get("start_string", "")
+    if not start_string:
+        start_string_template = "<!-- {} -->\n" if markdown_file else ".. {}\n"
+        start_string = start_string_template.format("towncrier release notes start")
+    parsed_data["start_string"] = start_string
 
-    if not os.path.exists(template):
-        raise ConfigError(
-            f"The template file '{template}' does not exist.",
-            failing_option="template",
-        )
-
-    return Config(
-        package=config.get("package", ""),
-        package_dir=config.get("package_dir", "."),
-        single_file=single_file,
-        filename=config.get("filename", "NEWS.rst"),
-        directory=config.get("directory"),
-        version=config.get("version"),
-        name=config.get("name"),
-        sections=sections,
-        types=types,
-        template=template,
-        start_string=config.get("start_string", _start_string),
-        title_format=config.get("title_format", _title_format),
-        issue_format=config.get("issue_format"),
-        underlines=config.get("underlines", _underlines),
-        wrap=wrap,
-        all_bullets=all_bullets,
-        orphan_prefix=config.get("orphan_prefix", "+"),
-    )
+    # Return the parsed config.
+    return Config(**parsed_data)
diff --git a/src/towncrier/build.py b/src/towncrier/build.py
index b5477f72..bb016443 100644
--- a/src/towncrier/build.py
+++ b/src/towncrier/build.py
@@ -25,6 +25,12 @@
 from ._writer import append_to_newsfile
 
 
+if sys.version_info < (3, 10):
+    import importlib_resources as resources
+else:
+    from importlib import resources
+
+
 def _get_date() -> str:
     return date.today().isoformat()
 
@@ -146,8 +152,11 @@ def __main(
     to_err = draft
 
     click.echo("Loading template...", err=to_err)
-    with open(config.template, "rb") as tmpl:
-        template = tmpl.read().decode("utf8")
+    if isinstance(config.template, tuple):
+        template = resources.read_text(*config.template)
+    else:
+        with open(config.template, encoding="utf-8") as tmpl:
+            template = tmpl.read()
 
     click.echo("Finding news fragments...", err=to_err)
 
diff --git a/src/towncrier/newsfragments/483.feature b/src/towncrier/newsfragments/483.feature
new file mode 100644
index 00000000..3397d34b
--- /dev/null
+++ b/src/towncrier/newsfragments/483.feature
@@ -0,0 +1,3 @@
+Provide a default Markdown template if the configured filename ends with ``.md``.
+
+The Markdown template uses the same rendered format as the default *reStructuredText* template, but with a Markdown syntax.
diff --git a/src/towncrier/templates/default.md b/src/towncrier/templates/default.md
new file mode 100644
index 00000000..45f7a395
--- /dev/null
+++ b/src/towncrier/templates/default.md
@@ -0,0 +1,61 @@
+{% if render_title %}
+{% if versiondata.name %}
+# {{ versiondata.name }} {{ versiondata.version }} ({{ versiondata.date }})
+{% else %}
+{{ versiondata.version }} ({{ versiondata.date }})
+{% endif %}
+{% endif %}
+{% for section, _ in sections.items() %}
+{% if section %}
+
+## {{section}}
+{% endif %}
+
+{% if sections[section] %}
+{% for category, val in definitions.items() if category in sections[section] %}
+### {{ definitions[category]['name'] }}
+
+{% if definitions[category]['showcontent'] %}
+{% for text, values in sections[section][category].items() %}
+- {{ text }}
+{%- if values %}
+{% if "\n  - " in text or '\n  * ' in text %}
+
+
+  (
+{%- else %}
+ (
+{%- endif -%}
+{%- for issue in values %}
+{{ issue.split(": ", 1)[0] }}{% if not loop.last %}, {% endif %}
+{%- endfor %}
+)
+{% else %}
+
+{% endif %}
+{% endfor %}
+
+{% else %}
+- {% for issue in sections[section][category][''] %}
+{{ issue.split(": ", 1)[0] }}{% if not loop.last %}, {% endif %}
+{% endfor %}
+
+
+{% endif %}
+{% if issues_by_category[section][category] and "]: " in issues_by_category[section][category][0] %}
+{% for issue in issues_by_category[section][category] %}
+{{ issue }}
+{% endfor %}
+
+{% endif %}
+{% if sections[section][category]|length == 0 %}
+No significant changes.
+
+{% else %}
+{% endif %}
+{% endfor %}
+{% else %}
+No significant changes.
+
+{% endif %}
+{% endfor %}
diff --git a/src/towncrier/test/helpers.py b/src/towncrier/test/helpers.py
index e86a9fcd..ddc9ffc6 100644
--- a/src/towncrier/test/helpers.py
+++ b/src/towncrier/test/helpers.py
@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import sys
+import textwrap
 
 from functools import wraps
 from pathlib import Path
@@ -19,12 +20,14 @@ def read(filename: str | Path) -> str:
     return Path(filename).read_text()
 
 
-def write(path: str | Path, contents: str) -> None:
+def write(path: str | Path, contents: str, dedent: bool = False) -> None:
     """
     Create a file with given contents including any missing parent directories
     """
     p = Path(path)
     p.parent.mkdir(parents=True, exist_ok=True)
+    if dedent:
+        contents = textwrap.dedent(contents)
     p.write_text(contents)
 
 
diff --git a/src/towncrier/test/test_build.py b/src/towncrier/test/test_build.py
index 0a76d916..4dbc07ce 100644
--- a/src/towncrier/test/test_build.py
+++ b/src/towncrier/test/test_build.py
@@ -15,7 +15,7 @@
 
 from .._shell import cli
 from ..build import _main
-from .helpers import read, setup_simple_project, with_isolated_runner
+from .helpers import read, setup_simple_project, with_isolated_runner, write
 
 
 class TestCli(TestCase):
@@ -1080,6 +1080,7 @@ def test_title_format_false(self):
                     "20-01-2001",
                     "--draft",
                 ],
+                catch_exceptions=False,
             )
 
         expected_output = dedent(
@@ -1162,6 +1163,103 @@ def test_start_string(self):
 
         self.assertEqual(expected_output, output)
 
+    @with_isolated_runner
+    def test_default_start_string(self, runner):
+        """
+        The default start string is ``.. towncrier release notes start``.
+        """
+        setup_simple_project()
+
+        write("foo/newsfragments/123.feature", "Adds levitation")
+        write(
+            "NEWS.rst",
+            contents="""
+                a line
+
+                another
+
+                .. towncrier release notes start
+
+                a footer!
+            """,
+            dedent=True,
+        )
+
+        result = runner.invoke(_main, ["--date", "01-01-2001"], catch_exceptions=False)
+        self.assertEqual(0, result.exit_code, result.output)
+        output = read("NEWS.rst")
+
+        expected_output = dedent(
+            """
+            a line
+
+            another
+
+            .. towncrier release notes start
+
+            Foo 1.2.3 (01-01-2001)
+            ======================
+
+            Features
+            --------
+
+            - Adds levitation (#123)
+
+
+            a footer!
+            """
+        )
+
+        self.assertEqual(expected_output, output)
+
+    @with_isolated_runner
+    def test_default_start_string_markdown(self, runner):
+        """
+        The default start string is ``<!-- towncrier release notes start -->`` for
+        Markdown.
+        """
+        setup_simple_project(extra_config='filename = "NEWS.md"')
+
+        write("foo/newsfragments/123.feature", "Adds levitation")
+        write(
+            "NEWS.md",
+            contents="""
+                a line
+
+                another
+
+                <!-- towncrier release notes start -->
+
+                a footer!
+            """,
+            dedent=True,
+        )
+
+        result = runner.invoke(_main, ["--date", "01-01-2001"], catch_exceptions=False)
+        self.assertEqual(0, result.exit_code, result.output)
+        output = read("NEWS.md")
+
+        expected_output = dedent(
+            """
+            a line
+
+            another
+
+            <!-- towncrier release notes start -->
+
+            # Foo 1.2.3 (01-01-2001)
+
+            ### Features
+
+            - Adds levitation (#123)
+
+
+            a footer!
+            """
+        )
+
+        self.assertEqual(expected_output, output)
+
     def test_with_topline_and_template_and_draft(self):
         """
         Spacing is proper when drafting with a topline and a template.
diff --git a/src/towncrier/test/test_format.py b/src/towncrier/test/test_format.py
index 241cbb8b..61e6d88c 100644
--- a/src/towncrier/test/test_format.py
+++ b/src/towncrier/test/test_format.py
@@ -2,8 +2,6 @@
 # See LICENSE for details.
 
 
-from collections import OrderedDict
-
 from twisted.trial.unittest import TestCase
 
 from .._builder import render_fragments, split_fragments
@@ -11,6 +9,8 @@
 
 
 class FormatterTests(TestCase):
+    maxDiff = None
+
     def test_split(self):
         fragments = {
             "": {
@@ -38,13 +38,11 @@ def test_split(self):
             },
         }
 
-        definitions = OrderedDict(
-            [
-                ("feature", {"name": "Features", "showcontent": True}),
-                ("bugfix", {"name": "Bugfixes", "showcontent": True}),
-                ("misc", {"name": "Misc", "showcontent": False}),
-            ]
-        )
+        definitions = {
+            "feature": {"name": "Features", "showcontent": True},
+            "bugfix": {"name": "Bugfixes", "showcontent": True},
+            "misc": {"name": "Misc", "showcontent": False},
+        }
 
         output = split_fragments(fragments, definitions)
 
@@ -55,36 +53,29 @@ def test_basic(self):
         Basic functionality -- getting a bunch of news fragments and formatting
         them into a rST file -- works.
         """
-        fragments = OrderedDict(
-            [
-                (
-                    "",
-                    {
-                        # asciibetical sorting will do 1, 142, 9
-                        # we want 1, 9, 142 instead
-                        ("142", "misc", 0): "",
-                        ("1", "misc", 0): "",
-                        ("9", "misc", 0): "",
-                        ("bar", "misc", 0): "",
-                        ("4", "feature", 0): "Stuff!",
-                        ("2", "feature", 0): "Foo added.",
-                        ("72", "feature", 0): "Foo added.",
-                        ("9", "feature", 0): "Foo added.",
-                        ("baz", "feature", 0): "Fun!",
-                    },
-                ),
-                ("Names", {}),
-                ("Web", {("3", "bugfix", 0): "Web fixed."}),
-            ]
-        )
+        fragments = {
+            "": {
+                # asciibetical sorting will do 1, 142, 9
+                # we want 1, 9, 142 instead
+                ("142", "misc", 0): "",
+                ("1", "misc", 0): "",
+                ("9", "misc", 0): "",
+                ("bar", "misc", 0): "",
+                ("4", "feature", 0): "Stuff!",
+                ("2", "feature", 0): "Foo added.",
+                ("72", "feature", 0): "Foo added.",
+                ("9", "feature", 0): "Foo added.",
+                ("baz", "feature", 0): "Fun!",
+            },
+            "Names": {},
+            "Web": {("3", "bugfix", 0): "Web fixed."},
+        }
 
-        definitions = OrderedDict(
-            [
-                ("feature", {"name": "Features", "showcontent": True}),
-                ("bugfix", {"name": "Bugfixes", "showcontent": True}),
-                ("misc", {"name": "Misc", "showcontent": False}),
-            ]
-        )
+        definitions = {
+            "feature": {"name": "Features", "showcontent": True},
+            "bugfix": {"name": "Bugfixes", "showcontent": True},
+            "misc": {"name": "Misc", "showcontent": False},
+        }
 
         expected_output = """MyProject 1.0 (never)
 =====================
@@ -176,6 +167,144 @@ def test_basic(self):
         )
         self.assertEqual(output, expected_output_weird_underlines)
 
+    def test_markdown(self):
+        """
+        Check formating of default markdown template.
+        """
+        fragments = {
+            "": {
+                # asciibetical sorting will do 1, 142, 9
+                # we want 1, 9, 142 instead
+                ("142", "misc", 0): "",
+                ("1", "misc", 0): "",
+                ("9", "misc", 0): "",
+                ("bar", "misc", 0): "",
+                ("4", "feature", 0): "Stuff!",
+                ("2", "feature", 0): "Foo added.",
+                ("72", "feature", 0): "Foo added.",
+                ("9", "feature", 0): "Foo added.",
+                ("3", "feature", 0): "Multi-line\nhere",
+                ("baz", "feature", 0): "Fun!",
+            },
+            "Names": {},
+            "Web": {
+                ("3", "bugfix", 0): "Web fixed.",
+                ("2", "bugfix", 0): "Multi-line bulleted\n- fix\n- here",
+            },
+        }
+
+        definitions = {
+            "feature": {"name": "Features", "showcontent": True},
+            "bugfix": {"name": "Bugfixes", "showcontent": True},
+            "misc": {"name": "Misc", "showcontent": False},
+        }
+
+        expected_output = """# MyProject 1.0 (never)
+
+### Features
+
+- Fun! (baz)
+- Foo added. (#2, #9, #72)
+- Multi-line
+  here (#3)
+- Stuff! (#4)
+
+### Misc
+
+- bar, #1, #9, #142
+
+
+## Names
+
+No significant changes.
+
+
+## Web
+
+### Bugfixes
+
+- Multi-line bulleted
+  - fix
+  - here
+
+  (#2)
+- Web fixed. (#3)
+"""
+
+        template = read_pkg_resource("templates/default.md")
+
+        fragments = split_fragments(fragments, definitions)
+        output = render_fragments(
+            template,
+            None,
+            fragments,
+            definitions,
+            ["-", "~"],
+            wrap=True,
+            versiondata={"name": "MyProject", "version": "1.0", "date": "never"},
+        )
+        self.assertEqual(output, expected_output)
+
+        # Also test with custom issue format
+        expected_output = """# MyProject 1.0 (never)
+
+### Features
+
+- Fun! ([baz])
+- Foo added. ([2], [9], [72])
+- Multi-line
+  here ([3])
+- Stuff! ([4])
+
+[baz]: https://github.com/twisted/towncrier/issues/baz
+[2]: https://github.com/twisted/towncrier/issues/2
+[3]: https://github.com/twisted/towncrier/issues/3
+[4]: https://github.com/twisted/towncrier/issues/4
+[9]: https://github.com/twisted/towncrier/issues/9
+[72]: https://github.com/twisted/towncrier/issues/72
+
+### Misc
+
+- [bar], [1], [9], [142]
+
+[bar]: https://github.com/twisted/towncrier/issues/bar
+[1]: https://github.com/twisted/towncrier/issues/1
+[9]: https://github.com/twisted/towncrier/issues/9
+[142]: https://github.com/twisted/towncrier/issues/142
+
+
+## Names
+
+No significant changes.
+
+
+## Web
+
+### Bugfixes
+
+- Multi-line bulleted
+  - fix
+  - here
+
+  ([2])
+- Web fixed. ([3])
+
+[2]: https://github.com/twisted/towncrier/issues/2
+[3]: https://github.com/twisted/towncrier/issues/3
+"""
+
+        output = render_fragments(
+            template,
+            "[{issue}]: https://github.com/twisted/towncrier/issues/{issue}",
+            fragments,
+            definitions,
+            ["-", "~"],
+            wrap=True,
+            versiondata={"name": "MyProject", "version": "1.0", "date": "never"},
+        )
+
+        self.assertEqual(output, expected_output)
+
     def test_issue_format(self):
         """
         issue_format option can be used to format issue text.
@@ -194,7 +323,7 @@ def test_issue_format(self):
             }
         }
 
-        definitions = OrderedDict([("misc", {"name": "Misc", "showcontent": False})])
+        definitions = {"misc": {"name": "Misc", "showcontent": False}}
 
         expected_output = """MyProject 1.0 (never)
 =====================
@@ -240,9 +369,7 @@ def test_line_wrapping(self):
             }
         }
 
-        definitions = OrderedDict(
-            [("feature", {"name": "Features", "showcontent": True})]
-        )
+        definitions = {"feature": {"name": "Features", "showcontent": True}}
 
         expected_output = """MyProject 1.0 (never)
 =====================
@@ -295,9 +422,7 @@ def test_line_wrapping_disabled(self):
             }
         }
 
-        definitions = OrderedDict(
-            [("feature", {"name": "Features", "showcontent": True})]
-        )
+        definitions = {"feature": {"name": "Features", "showcontent": True}}
 
         expected_output = """MyProject 1.0 (never)
 =====================
diff --git a/src/towncrier/test/test_project.py b/src/towncrier/test/test_project.py
index 4b420648..c41c3a55 100644
--- a/src/towncrier/test/test_project.py
+++ b/src/towncrier/test/test_project.py
@@ -25,7 +25,6 @@ def test_str(self):
         A str __version__ will be picked up.
         """
         temp = self.mktemp()
-        os.makedirs(temp)
         os.makedirs(os.path.join(temp, "mytestproj"))
 
         with open(os.path.join(temp, "mytestproj", "__init__.py"), "w") as f:
@@ -39,7 +38,6 @@ def test_tuple(self):
         A tuple __version__ will be picked up.
         """
         temp = self.mktemp()
-        os.makedirs(temp)
         os.makedirs(os.path.join(temp, "mytestproja"))
 
         with open(os.path.join(temp, "mytestproja", "__init__.py"), "w") as f:
@@ -109,7 +107,6 @@ def test_unknown_type(self):
         A __version__ of unknown type will lead to an exception.
         """
         temp = self.mktemp()
-        os.makedirs(temp)
         os.makedirs(os.path.join(temp, "mytestprojb"))
 
         with open(os.path.join(temp, "mytestprojb", "__init__.py"), "w") as f:
@@ -133,14 +130,12 @@ def test_already_installed_import(self):
         project_name = "mytestproj_already_installed_import"
 
         temp = self.mktemp()
-        os.makedirs(temp)
         os.makedirs(os.path.join(temp, project_name))
 
         with open(os.path.join(temp, project_name, "__init__.py"), "w") as f:
             f.write("__version__ = (1, 3, 12)")
 
         sys_path_temp = self.mktemp()
-        os.makedirs(sys_path_temp)
         os.makedirs(os.path.join(sys_path_temp, project_name))
 
         with open(os.path.join(sys_path_temp, project_name, "__init__.py"), "w") as f:
@@ -161,7 +156,6 @@ def test_installed_package_found_when_no_source_present(self):
         project_name = "mytestproj_only_installed"
 
         sys_path_temp = self.mktemp()
-        os.makedirs(sys_path_temp)
         os.makedirs(os.path.join(sys_path_temp, project_name))
 
         with open(os.path.join(sys_path_temp, project_name, "__init__.py"), "w") as f:
diff --git a/src/towncrier/test/test_settings.py b/src/towncrier/test/test_settings.py
index c954e738..3c2dca94 100644
--- a/src/towncrier/test/test_settings.py
+++ b/src/towncrier/test/test_settings.py
@@ -1,59 +1,128 @@
 # Copyright (c) Amber Brown, 2015
 # See LICENSE for details.
 
-import collections as clt
 import os
-import textwrap
-
-from textwrap import dedent
 
 from twisted.trial.unittest import TestCase
 
 from .._settings import ConfigError, load_config
+from .helpers import write
 
 
 class TomlSettingsTests(TestCase):
+    def mktemp_project(
+        self, *, pyproject_toml: str = "", towncrier_toml: str = ""
+    ) -> str:
+        """
+        Create a temporary directory with a pyproject.toml file in it.
+        """
+        project_dir = self.mktemp()
+        os.makedirs(project_dir)
+
+        if pyproject_toml:
+            write(
+                os.path.join(project_dir, "pyproject.toml"),
+                pyproject_toml,
+                dedent=True,
+            )
+
+        if towncrier_toml:
+            write(
+                os.path.join(project_dir, "towncrier.toml"),
+                towncrier_toml,
+                dedent=True,
+            )
+
+        return project_dir
+
     def test_base(self):
         """
         Test a "base config".
         """
-        temp = self.mktemp()
-        os.makedirs(temp)
-
-        with open(os.path.join(temp, "pyproject.toml"), "w") as f:
-            f.write(
-                """[tool.towncrier]
-package = "foobar"
-orphan_prefix = "~"
-"""
-            )
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
+                [tool.towncrier]
+                package = "foobar"
+                orphan_prefix = "~"
+            """
+        )
 
-        config = load_config(temp)
+        config = load_config(project_dir)
         self.assertEqual(config.package, "foobar")
         self.assertEqual(config.package_dir, ".")
         self.assertEqual(config.filename, "NEWS.rst")
-        self.assertEqual(config.underlines, ["=", "-", "~"])
+        self.assertEqual(config.underlines, ("=", "-", "~"))
         self.assertEqual(config.orphan_prefix, "~")
 
+    def test_markdown(self):
+        """
+        If the filename references an .md file and the builtin template doesn't have an
+        extension, add .md rather than .rst.
+        """
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
+                [tool.towncrier]
+                package = "foobar"
+                filename = "NEWS.md"
+            """
+        )
+
+        config = load_config(project_dir)
+
+        self.assertEqual(config.filename, "NEWS.md")
+
+        self.assertEqual(config.template, ("towncrier.templates", "default.md"))
+
+    def test_explicit_template_extension(self):
+        """
+        If the filename references an .md file and the builtin template has an
+        extension, don't change it.
+        """
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
+                [tool.towncrier]
+                package = "foobar"
+                filename = "NEWS.md"
+                template = "towncrier:default.rst"
+            """
+        )
+
+        config = load_config(project_dir)
+
+        self.assertEqual(config.filename, "NEWS.md")
+        self.assertEqual(config.template, ("towncrier.templates", "default.rst"))
+
+    def test_template_extended(self):
+        """
+        The template can be any package and resource, and although we look for a
+        resource's 'templates' package, it could also be in the specified resource
+        directly.
+        """
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
+                [tool.towncrier]
+                package = "foobar"
+                template = "towncrier.templates:default.rst"
+            """
+        )
+
+        config = load_config(project_dir)
+
+        self.assertEqual(config.template, ("towncrier.templates", "default.rst"))
+
     def test_missing(self):
         """
         If the config file doesn't have the correct toml key, we error.
         """
-        temp = self.mktemp()
-        os.makedirs(temp)
-
-        with open(os.path.join(temp, "pyproject.toml"), "w") as f:
-            f.write(
-                dedent(
-                    """
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
                 [something.else]
                 blah='baz'
-                """
-                )
-            )
+            """
+        )
 
         with self.assertRaises(ConfigError) as e:
-            load_config(temp)
+            load_config(project_dir)
 
         self.assertEqual(e.exception.failing_option, "all")
 
@@ -61,21 +130,15 @@ def test_incorrect_single_file(self):
         """
         single_file must be a bool.
         """
-        temp = self.mktemp()
-        os.makedirs(temp)
-
-        with open(os.path.join(temp, "pyproject.toml"), "w") as f:
-            f.write(
-                dedent(
-                    """
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
                 [tool.towncrier]
                 single_file = "a"
-                """
-                )
-            )
+            """
+        )
 
         with self.assertRaises(ConfigError) as e:
-            load_config(temp)
+            load_config(project_dir)
 
         self.assertEqual(e.exception.failing_option, "single_file")
 
@@ -83,21 +146,15 @@ def test_incorrect_all_bullets(self):
         """
         all_bullets must be a bool.
         """
-        temp = self.mktemp()
-        os.makedirs(temp)
-
-        with open(os.path.join(temp, "pyproject.toml"), "w") as f:
-            f.write(
-                dedent(
-                    """
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
                 [tool.towncrier]
                 all_bullets = "a"
-                """
-                )
-            )
+            """
+        )
 
         with self.assertRaises(ConfigError) as e:
-            load_config(temp)
+            load_config(project_dir)
 
         self.assertEqual(e.exception.failing_option, "all_bullets")
 
@@ -105,21 +162,15 @@ def test_mistype_singlefile(self):
         """
         singlefile is not accepted, single_file is.
         """
-        temp = self.mktemp()
-        os.makedirs(temp)
-
-        with open(os.path.join(temp, "pyproject.toml"), "w") as f:
-            f.write(
-                dedent(
-                    """
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
                 [tool.towncrier]
                 singlefile = "a"
-                """
-                )
-            )
+            """
+        )
 
         with self.assertRaises(ConfigError) as e:
-            load_config(temp)
+            load_config(project_dir)
 
         self.assertEqual(e.exception.failing_option, "singlefile")
 
@@ -127,56 +178,38 @@ def test_towncrier_toml_preferred(self):
         """
         Towncrier prefers the towncrier.toml for autodetect over pyproject.toml.
         """
-        temp = self.mktemp()
-        os.makedirs(temp)
-
-        with open(os.path.join(temp, "towncrier.toml"), "w") as f:
-            f.write(
-                dedent(
-                    """
+        project_dir = self.mktemp_project(
+            towncrier_toml="""
                 [tool.towncrier]
                 package = "a"
-                """
-                )
-            )
-
-        with open(os.path.join(temp, "pyproject.toml"), "w") as f:
-            f.write(
-                dedent(
-                    """
+            """,
+            pyproject_toml="""
                 [tool.towncrier]
                 package = "b"
-                """
-                )
-            )
+            """,
+        )
 
-        config = load_config(temp)
+        config = load_config(project_dir)
         self.assertEqual(config.package, "a")
 
     def test_missing_template(self):
         """
         Towncrier will raise an exception saying when it can't find a template.
         """
-        temp = self.mktemp()
-        os.makedirs(temp)
-
-        with open(os.path.join(temp, "towncrier.toml"), "w") as f:
-            f.write(
-                dedent(
-                    """
+        project_dir = self.mktemp_project(
+            towncrier_toml="""
                 [tool.towncrier]
                 template = "foo.rst"
-                """
-                )
-            )
+            """
+        )
 
         with self.assertRaises(ConfigError) as e:
-            load_config(temp)
+            load_config(project_dir)
 
         self.assertEqual(
             str(e.exception),
             "The template file '{}' does not exist.".format(
-                os.path.normpath(os.path.join(temp, "foo.rst")),
+                os.path.normpath(os.path.join(project_dir, "foo.rst")),
             ),
         )
 
@@ -185,24 +218,18 @@ def test_missing_template_in_towncrier(self):
         Towncrier will raise an exception saying when it can't find a template
         from the Towncrier templates.
         """
-        temp = self.mktemp()
-        os.makedirs(temp)
-
-        with open(os.path.join(temp, "towncrier.toml"), "w") as f:
-            f.write(
-                dedent(
-                    """
+        project_dir = self.mktemp_project(
+            towncrier_toml="""
                 [tool.towncrier]
                 template = "towncrier:foo"
-                """
-                )
-            )
+            """
+        )
 
         with self.assertRaises(ConfigError) as e:
-            load_config(temp)
+            load_config(project_dir)
 
         self.assertEqual(
-            str(e.exception), "Towncrier does not have a template named 'foo'."
+            str(e.exception), "'towncrier' does not have a template named 'foo.rst'."
         )
 
     def test_custom_types_as_tables_array_deprecated(self):
@@ -214,20 +241,22 @@ def test_custom_types_as_tables_array_deprecated(self):
         This functionality is considered deprecated, but we continue
         to support it to keep backward compatibility.
         """
-        toml_content = """
-        [tool.towncrier]
-        package = "foobar"
-        [[tool.towncrier.type]]
-        directory="foo"
-        name="Foo"
-        showcontent=false
-
-        [[tool.towncrier.type]]
-        directory="spam"
-        name="Spam"
-        showcontent=true
-        """
-        toml_content = textwrap.dedent(toml_content)
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
+                [tool.towncrier]
+                package = "foobar"
+                [[tool.towncrier.type]]
+                directory="foo"
+                name="Foo"
+                showcontent=false
+
+                [[tool.towncrier.type]]
+                directory="spam"
+                name="Spam"
+                showcontent=true
+            """
+        )
+        config = load_config(project_dir)
         expected = [
             (
                 "foo",
@@ -244,10 +273,7 @@ def test_custom_types_as_tables_array_deprecated(self):
                 },
             ),
         ]
-        expected = clt.OrderedDict(expected)
-        config = self.load_config_from_string(
-            toml_content,
-        )
+        expected = dict(expected)
         actual = config.types
         self.assertDictEqual(expected, actual)
 
@@ -256,60 +282,32 @@ def test_custom_types_as_tables(self):
         Custom fragment categories can be defined inside
         the toml config file using tables.
         """
-        test_project_path = self.mktemp()
-        os.makedirs(test_project_path)
-        toml_content = """
-        [tool.towncrier]
-        package = "foobar"
-        [tool.towncrier.fragment.feat]
-        ignored_field="Bazz"
-        [tool.towncrier.fragment.fix]
-        [tool.towncrier.fragment.chore]
-        name = "Other Tasks"
-        showcontent = false
-        """
-        toml_content = textwrap.dedent(toml_content)
-        expected = [
-            (
-                "chore",
-                {
-                    "name": "Other Tasks",
-                    "showcontent": False,
-                },
-            ),
-            (
-                "feat",
-                {
-                    "name": "Feat",
-                    "showcontent": True,
-                },
-            ),
-            (
-                "fix",
-                {
-                    "name": "Fix",
-                    "showcontent": True,
-                },
-            ),
-        ]
-
-        expected = clt.OrderedDict(expected)
-        config = self.load_config_from_string(
-            toml_content,
+        project_dir = self.mktemp_project(
+            pyproject_toml="""
+                [tool.towncrier]
+                package = "foobar"
+                [tool.towncrier.fragment.feat]
+                ignored_field="Bazz"
+                [tool.towncrier.fragment.fix]
+                [tool.towncrier.fragment.chore]
+                name = "Other Tasks"
+                showcontent = false
+            """
         )
+        config = load_config(project_dir)
+        expected = {
+            "chore": {
+                "name": "Other Tasks",
+                "showcontent": False,
+            },
+            "feat": {
+                "name": "Feat",
+                "showcontent": True,
+            },
+            "fix": {
+                "name": "Fix",
+                "showcontent": True,
+            },
+        }
         actual = config.types
         self.assertDictEqual(expected, actual)
-
-    def load_config_from_string(self, toml_content):
-        """Load configuration from a string.
-
-        Given a string following toml syntax,
-        obtain the towncrier configuration.
-        """
-        test_project_path = self.mktemp()
-        os.makedirs(test_project_path)
-        toml_path = os.path.join(test_project_path, "pyproject.toml")
-        with open(toml_path, "w") as f:
-            f.write(toml_content)
-        config = load_config(test_project_path)
-        return config
diff --git a/src/towncrier/test/test_write.py b/src/towncrier/test/test_write.py
index 83443f20..563fde6f 100644
--- a/src/towncrier/test/test_write.py
+++ b/src/towncrier/test/test_write.py
@@ -3,7 +3,6 @@
 
 import os
 
-from collections import OrderedDict
 from pathlib import Path
 from textwrap import dedent
 
@@ -13,40 +12,31 @@
 from .._builder import render_fragments, split_fragments
 from .._writer import append_to_newsfile
 from ..build import _main
-from .helpers import read_pkg_resource
+from .helpers import read_pkg_resource, write
 
 
 class WritingTests(TestCase):
     maxDiff = None
 
     def test_append_at_top(self):
-        fragments = OrderedDict(
-            [
-                (
-                    "",
-                    OrderedDict(
-                        [
-                            (("142", "misc", 0), ""),
-                            (("1", "misc", 0), ""),
-                            (("4", "feature", 0), "Stuff!"),
-                            (("4", "feature", 1), "Second Stuff!"),
-                            (("2", "feature", 0), "Foo added."),
-                            (("72", "feature", 0), "Foo added."),
-                        ]
-                    ),
-                ),
-                ("Names", {}),
-                ("Web", {("3", "bugfix", 0): "Web fixed."}),
-            ]
-        )
-
-        definitions = OrderedDict(
-            [
-                ("feature", {"name": "Features", "showcontent": True}),
-                ("bugfix", {"name": "Bugfixes", "showcontent": True}),
-                ("misc", {"name": "Misc", "showcontent": False}),
-            ]
-        )
+        fragments = {
+            "": {
+                ("142", "misc", 0): "",
+                ("1", "misc", 0): "",
+                ("4", "feature", 0): "Stuff!",
+                ("4", "feature", 1): "Second Stuff!",
+                ("2", "feature", 0): "Foo added.",
+                ("72", "feature", 0): "Foo added.",
+            },
+            "Names": {},
+            "Web": {("3", "bugfix", 0): "Web fixed."},
+        }
+
+        definitions = {
+            "feature": {"name": "Features", "showcontent": True},
+            "bugfix": {"name": "Bugfixes", "showcontent": True},
+            "misc": {"name": "Misc", "showcontent": False},
+        }
 
         expected_output = """MyProject 1.0 (never)
 =====================
@@ -84,7 +74,7 @@ def test_append_at_top(self):
 """
 
         tempdir = self.mktemp()
-        os.mkdir(tempdir)
+        os.makedirs(tempdir)
 
         with open(os.path.join(tempdir, "NEWS.rst"), "w") as f:
             f.write("Old text.\n")
@@ -120,31 +110,24 @@ def test_append_at_top_with_hint(self):
         If there is a comment with C{.. towncrier release notes start},
         towncrier will add the version notes after it.
         """
-        fragments = OrderedDict(
-            [
-                (
-                    "",
-                    {
-                        ("142", "misc", 0): "",
-                        ("1", "misc", 0): "",
-                        ("4", "feature", 0): "Stuff!",
-                        ("2", "feature", 0): "Foo added.",
-                        ("72", "feature", 0): "Foo added.",
-                        ("99", "feature", 0): "Foo! " * 100,
-                    },
-                ),
-                ("Names", {}),
-                ("Web", {("3", "bugfix", 0): "Web fixed."}),
-            ]
-        )
-
-        definitions = OrderedDict(
-            [
-                ("feature", {"name": "Features", "showcontent": True}),
-                ("bugfix", {"name": "Bugfixes", "showcontent": True}),
-                ("misc", {"name": "Misc", "showcontent": False}),
-            ]
-        )
+        fragments = {
+            "": {
+                ("142", "misc", 0): "",
+                ("1", "misc", 0): "",
+                ("4", "feature", 0): "Stuff!",
+                ("2", "feature", 0): "Foo added.",
+                ("72", "feature", 0): "Foo added.",
+                ("99", "feature", 0): "Foo! " * 100,
+            },
+            "Names": {},
+            "Web": {("3", "bugfix", 0): "Web fixed."},
+        }
+
+        definitions = {
+            "feature": {"name": "Features", "showcontent": True},
+            "bugfix": {"name": "Bugfixes", "showcontent": True},
+            "misc": {"name": "Misc", "showcontent": False},
+        }
 
         expected_output = """Hello there! Here is some info.
 
@@ -192,13 +175,16 @@ def test_append_at_top_with_hint(self):
 """
 
         tempdir = self.mktemp()
-        os.mkdir(tempdir)
-
-        with open(os.path.join(tempdir, "NEWS.rst"), "w") as f:
-            f.write(
-                "Hello there! Here is some info.\n\n"
-                ".. towncrier release notes start\nOld text.\n"
-            )
+        write(
+            os.path.join(tempdir, "NEWS.rst"),
+            contents="""\
+                Hello there! Here is some info.
+
+                .. towncrier release notes start
+                Old text.
+            """,
+            dedent=True,
+        )
 
         fragments = split_fragments(fragments, definitions)
 
@@ -232,7 +218,7 @@ def test_multiple_file_no_start_string(self):
         the start of the file.
         """
         tempdir = self.mktemp()
-        os.mkdir(tempdir)
+        os.makedirs(tempdir)
 
         definitions = {}
         fragments = split_fragments(fragments={}, definitions=definitions)