How to run one specific pytest test from the command line

Your test suite has 800 tests, one of them is failing, and you don’t want to run the other 799 while you iterate on the fix. Here are six ways to narrow pytest down.

By File #

pytest tests/test_users.py

By Test Function #

pytest tests/test_users.py::test_user_signup

Two colons separate the file from the identifier inside it. Not one - : is what fingers type first, and it fails silently (pytest collects nothing).

By Class and Method #

pytest tests/test_users.py::TestUserService::test_creates_user

By Parametrised Case #

If your test is @pytest.mark.parametrized, you can select one case by its ID:

pytest 'tests/test_users.py::test_user_signup[admin-True]'

Wrap in quotes because the brackets are shell metacharacters.

Not sure what the IDs look like? List them without running:

pytest tests/test_users.py --collect-only -q

By Keyword (across the whole suite) #

The -k flag matches against the test name as a Python expression, so and, or, and not work:

pytest -k signup
pytest -k 'signup and not admin'

Handy when you don’t remember which file the test is in.

By Marker #

If your tests have @pytest.mark.slow (or your own markers):

pytest -m slow
pytest -m 'not slow'

Just Re-Run What Failed #

pytest --last-failed
# or the shorthand
pytest --lf

Runs only tests that failed in the previous run. The complementary flag is --failed-first (--ff), which runs everything but puts the previously-failed tests first.

The Combo I Use Most #

pytest -x -k signup --lf

• -x stops at the first failure
• -k signup narrows to the tests you’re iterating on
• --lf prefers the ones that were already broken

Tight iteration loop.