diff --git a/stubs/invoke/@tests/test_cases/check_task.py b/stubs/invoke/@tests/test_cases/check_task.py
index 6ad7b8cd0..06b96ad79 100644
--- a/stubs/invoke/@tests/test_cases/check_task.py
+++ b/stubs/invoke/@tests/test_cases/check_task.py
@@ -1,4 +1,5 @@
 # pyright: reportUnnecessaryTypeIgnoreComment=true
+from __future__ import annotations
 
 from invoke import Context, task
 
diff --git a/test_cases/README.md b/test_cases/README.md
index 4ff8949d1..b154c0d2e 100644
--- a/test_cases/README.md
+++ b/test_cases/README.md
@@ -79,20 +79,26 @@ is because the purpose of this folder is to test the implications of typeshed
 changes for end users, who will mainly be using `.py` files rather than `.pyi`
 files.
 
-Another difference to the rest of typeshed is that the test cases in this
-directory cannot always use modern syntax for type hints.
+Another difference to the rest of typeshed
+(which stems from the fact that the test-case files are all `.py` files
+rather than `.pyi` files)
+is that the test cases cannot always use modern syntax for type hints.
+While we can use `from __future__ import annotations` to enable the use of
+modern typing syntax wherever possible,
+type checkers may (correctly) emit errors if PEP 604 syntax or PEP 585 syntax
+is used in a runtime context on lower versions of Python. For example:
 
-For example, PEP 604
-syntax (unions with a pipe `|` operator) is new in Python 3.10. While this
-syntax can be used on older Python versions in a `.pyi` file, code using this
-syntax will fail at runtime on Python <=3.9. Since the test cases all use `.py`
-extensions, and since the tests need to pass on all Python versions >=3.7, PEP
-604 syntax cannot be used in a test case. Use `typing.Union` and
-`typing.Optional` instead.
+```python
+from __future__ import annotations
 
-PEP 585 syntax can also not be used in the `test_cases` directory. Use
-`typing.Tuple` instead of `tuple`, `typing.Callable` instead of
-`collections.abc.Callable`, and `typing.Match` instead of `re.Match` (etc.).
+from typing_extensions import assert_type
+
+x: str | int  # PEP 604 syntax: okay on Python >=3.7, due to __future__ annotations
+assert_type(x, str | int)  # Will fail at runtime on Python <3.10 (use typing.Union instead)
+
+y: dict[str, int]  # PEP 585 syntax: okay on Python >= 3.7, due to __future__ annotations
+assert_type(y, dict[str, int])  # Will fail at runtime on Python <3.9 (use typing.Dict instead)
+```
 
 ### Version-dependent tests
 
diff --git a/test_cases/stdlib/asyncio/check_coroutines.py b/test_cases/stdlib/asyncio/check_coroutines.py
index e2c3d19e6..cfee8de0e 100644
--- a/test_cases/stdlib/asyncio/check_coroutines.py
+++ b/test_cases/stdlib/asyncio/check_coroutines.py
@@ -1,13 +1,15 @@
+from __future__ import annotations
+
 from asyncio import iscoroutinefunction
 from collections.abc import Awaitable, Callable, Coroutine
-from typing import Any, Union
+from typing import Any
 from typing_extensions import assert_type
 
 
 def test_iscoroutinefunction(
     x: Callable[[str, int], Coroutine[str, int, bytes]],
     y: Callable[[str, int], Awaitable[bytes]],
-    z: Callable[[str, int], Union[str, Awaitable[bytes]]],
+    z: Callable[[str, int], str | Awaitable[bytes]],
     xx: object,
 ) -> None:
 
diff --git a/test_cases/stdlib/asyncio/check_gather.py b/test_cases/stdlib/asyncio/check_gather.py
index c64af0b11..6b7d3ce3d 100644
--- a/test_cases/stdlib/asyncio/check_gather.py
+++ b/test_cases/stdlib/asyncio/check_gather.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import asyncio
 from typing import Any, Awaitable, List, Tuple, Union
 from typing_extensions import assert_type
@@ -21,7 +23,7 @@ async def test_gather(awaitable1: Awaitable[int], awaitable2: Awaitable[str]) ->
     c = await asyncio.gather(awaitable1, awaitable2, awaitable1, awaitable1, awaitable1, awaitable1)
     assert_type(c, List[Any])
 
-    awaitables_list: List[Awaitable[int]] = [awaitable1]
+    awaitables_list: list[Awaitable[int]] = [awaitable1]
     d = await asyncio.gather(*awaitables_list)
     assert_type(d, List[Any])
 
diff --git a/test_cases/stdlib/builtins/check_dict.py b/test_cases/stdlib/builtins/check_dict.py
index 3ba6ce89a..2161bb264 100644
--- a/test_cases/stdlib/builtins/check_dict.py
+++ b/test_cases/stdlib/builtins/check_dict.py
@@ -1,4 +1,6 @@
-from typing import Dict, Generic, Iterable, Tuple, TypeVar
+from __future__ import annotations
+
+from typing import Dict, Generic, Iterable, TypeVar
 from typing_extensions import assert_type
 
 # These do follow `__init__` overloads order:
@@ -6,7 +8,7 @@ from typing_extensions import assert_type
 # mypy raises: 'Need type annotation for "bad"'
 # pyright is fine with it.
 # bad = dict()
-good: Dict[str, str] = dict()
+good: dict[str, str] = dict()
 assert_type(good, Dict[str, str])
 
 assert_type(dict(arg=1), Dict[str, int])
@@ -16,7 +18,7 @@ _VT = TypeVar("_VT")
 
 
 class KeysAndGetItem(Generic[_KT, _VT]):
-    data: Dict[_KT, _VT]
+    data: dict[_KT, _VT]
 
     def keys(self) -> Iterable[_KT]:
         return self.data.keys()
@@ -33,15 +35,15 @@ kt2: KeysAndGetItem[str, int] = KeysAndGetItem()
 assert_type(dict(kt2, arg=1), Dict[str, int])
 
 
-def test_iterable_tuple_overload(x: Iterable[Tuple[int, str]]) -> Dict[int, str]:
+def test_iterable_tuple_overload(x: Iterable[tuple[int, str]]) -> dict[int, str]:
     return dict(x)
 
 
-i1: Iterable[Tuple[int, str]] = [(1, "a"), (2, "b")]
+i1: Iterable[tuple[int, str]] = [(1, "a"), (2, "b")]
 test_iterable_tuple_overload(i1)
 dict(i1, arg="a")  # type: ignore
 
-i2: Iterable[Tuple[str, int]] = [("a", 1), ("b", 2)]
+i2: Iterable[tuple[str, int]] = [("a", 1), ("b", 2)]
 assert_type(dict(i2, arg=1), Dict[str, int])
 
 i3: Iterable[str] = ["a.b"]
diff --git a/test_cases/stdlib/builtins/check_iteration.py b/test_cases/stdlib/builtins/check_iteration.py
index 7195c8bcd..3d6096353 100644
--- a/test_cases/stdlib/builtins/check_iteration.py
+++ b/test_cases/stdlib/builtins/check_iteration.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from typing import Iterator
 from typing_extensions import assert_type
 
diff --git a/test_cases/stdlib/builtins/check_list.py b/test_cases/stdlib/builtins/check_list.py
index 793d7951b..4113f5c66 100644
--- a/test_cases/stdlib/builtins/check_list.py
+++ b/test_cases/stdlib/builtins/check_list.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from typing import List, Union
 from typing_extensions import assert_type
 
diff --git a/test_cases/stdlib/builtins/check_object.py b/test_cases/stdlib/builtins/check_object.py
index f2b56e4e4..60df1143f 100644
--- a/test_cases/stdlib/builtins/check_object.py
+++ b/test_cases/stdlib/builtins/check_object.py
@@ -1,9 +1,11 @@
-from typing import Any, Tuple, Union
+from __future__ import annotations
+
+from typing import Any
 
 
 # The following should pass without error (see #6661):
 class Diagnostic:
-    def __reduce__(self) -> Union[str, Tuple[Any, ...]]:
+    def __reduce__(self) -> str | tuple[Any, ...]:
         res = super().__reduce__()
         if isinstance(res, tuple) and len(res) >= 3:
             res[2]["_info"] = 42
diff --git a/test_cases/stdlib/builtins/check_pow.py b/test_cases/stdlib/builtins/check_pow.py
index c1ff7497e..da06229f4 100644
--- a/test_cases/stdlib/builtins/check_pow.py
+++ b/test_cases/stdlib/builtins/check_pow.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from decimal import Decimal
 from fractions import Fraction
 from typing import Any
diff --git a/test_cases/stdlib/builtins/check_sum.py b/test_cases/stdlib/builtins/check_sum.py
index 0f256fd2f..1228ff13b 100644
--- a/test_cases/stdlib/builtins/check_sum.py
+++ b/test_cases/stdlib/builtins/check_sum.py
@@ -1,26 +1,28 @@
+from __future__ import annotations
+
 from typing import Any, List, Union
 from typing_extensions import Literal, assert_type
 
 
 class Foo:
-    def __add__(self, other: Any) -> "Foo":
+    def __add__(self, other: Any) -> Foo:
         return Foo()
 
 
 class Bar:
-    def __radd__(self, other: Any) -> "Bar":
+    def __radd__(self, other: Any) -> Bar:
         return Bar()
 
 
 class Baz:
-    def __add__(self, other: Any) -> "Baz":
+    def __add__(self, other: Any) -> Baz:
         return Baz()
 
-    def __radd__(self, other: Any) -> "Baz":
+    def __radd__(self, other: Any) -> Baz:
         return Baz()
 
 
-literal_list: List[Literal[0, 1]] = [0, 1, 1]
+literal_list: list[Literal[0, 1]] = [0, 1, 1]
 
 assert_type(sum([2, 4]), int)
 assert_type(sum([3, 5], 4), int)
diff --git a/test_cases/stdlib/builtins/check_tuple.py b/test_cases/stdlib/builtins/check_tuple.py
index bf066feb2..bc0d8db28 100644
--- a/test_cases/stdlib/builtins/check_tuple.py
+++ b/test_cases/stdlib/builtins/check_tuple.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from typing import Tuple
 from typing_extensions import assert_type
 
diff --git a/test_cases/stdlib/check_codecs.py b/test_cases/stdlib/check_codecs.py
index 0657416bc..19e663cee 100644
--- a/test_cases/stdlib/check_codecs.py
+++ b/test_cases/stdlib/check_codecs.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import codecs
 from typing_extensions import assert_type
 
diff --git a/test_cases/stdlib/check_contextlib.py b/test_cases/stdlib/check_contextlib.py
index 26253326b..648661bca 100644
--- a/test_cases/stdlib/check_contextlib.py
+++ b/test_cases/stdlib/check_contextlib.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from contextlib import ExitStack
 from typing_extensions import assert_type
 
diff --git a/test_cases/stdlib/check_logging.py b/test_cases/stdlib/check_logging.py
index dd572538f..f51edd19d 100644
--- a/test_cases/stdlib/check_logging.py
+++ b/test_cases/stdlib/check_logging.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import logging
 from typing import Any
 
diff --git a/test_cases/stdlib/check_threading.py b/test_cases/stdlib/check_threading.py
index 9b67ac540..eddfc2549 100644
--- a/test_cases/stdlib/check_threading.py
+++ b/test_cases/stdlib/check_threading.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import _threading_local
 import threading
 
diff --git a/test_cases/stdlib/check_unittest.py b/test_cases/stdlib/check_unittest.py
index b8a16e7ef..7b0de9b30 100644
--- a/test_cases/stdlib/check_unittest.py
+++ b/test_cases/stdlib/check_unittest.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import unittest
 from datetime import datetime, timedelta
 from decimal import Decimal
@@ -58,14 +60,14 @@ class Eggs:
 
 
 class Ham:
-    def __lt__(self, other: "Ham") -> bool:
+    def __lt__(self, other: Ham) -> bool:
         if not isinstance(other, Ham):
             return NotImplemented
         return True
 
 
 class Bacon:
-    def __gt__(self, other: "Bacon") -> bool:
+    def __gt__(self, other: Bacon) -> bool:
         if not isinstance(other, Bacon):
             return NotImplemented
         return True
diff --git a/test_cases/stdlib/typing/check_all.py b/test_cases/stdlib/typing/check_all.py
index 3519912e1..9a5165383 100644
--- a/test_cases/stdlib/typing/check_all.py
+++ b/test_cases/stdlib/typing/check_all.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 # pyright: reportWildcardImportFromLibrary=false
 
 """
diff --git a/test_cases/stdlib/typing/check_pattern.py b/test_cases/stdlib/typing/check_pattern.py
index 69978c4aa..ec5c1c4f6 100644
--- a/test_cases/stdlib/typing/check_pattern.py
+++ b/test_cases/stdlib/typing/check_pattern.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from typing import Match, Optional, Pattern
 from typing_extensions import assert_type
 
diff --git a/tests/check_consistent.py b/tests/check_consistent.py
index 4b2931524..2a5e8d93a 100755
--- a/tests/check_consistent.py
+++ b/tests/check_consistent.py
@@ -85,9 +85,10 @@ def check_test_cases() -> None:
         bad_test_case_filename = 'Files in a `test_cases` directory must have names starting with "check_"; got "{}"'
         for file in testcase_dir.rglob("*.py"):
             assert file.stem.startswith("check_"), bad_test_case_filename.format(file)
+            with open(file, encoding="UTF-8") as f:
+                lines = {line.strip() for line in f}
+            assert "from __future__ import annotations" in lines, "Test-case files should use modern typing syntax where possible"
             if package_name != "stdlib":
-                with open(file, encoding="UTF-8") as f:
-                    lines = {line.strip() for line in f}
                 pyright_setting_not_enabled_msg = (
                     f'Third-party test-case file "{file}" must have '
                     f'"# pyright: reportUnnecessaryTypeIgnoreComment=true" '
@@ -205,7 +206,7 @@ def check_requirements() -> None:
     precommit_requirements = get_precommit_requirements()
     no_txt_entry_msg = "All pre-commit requirements must also be listed in `requirements-tests.txt` (missing {requirement!r})"
     for requirement, specifier in precommit_requirements.items():
-        assert requirement in requirements_txt_requirements, no_txt_entry_msg.format(requirement)
+        assert requirement in requirements_txt_requirements, no_txt_entry_msg.format(requirement=requirement)
         specifier_mismatch = (
             f'Specifier "{specifier}" for {requirement!r} in `.pre-commit-config.yaml` '
             f'does not match specifier "{requirements_txt_requirements[requirement]}" in `requirements-tests.txt`'