diff --git a/tests/e2e_integration_tests/action_masking/test_agents_use_action_masks.py b/tests/e2e_integration_tests/action_masking/test_agents_use_action_masks.py
index 745e280b..4260c605 100644
--- a/tests/e2e_integration_tests/action_masking/test_agents_use_action_masks.py
+++ b/tests/e2e_integration_tests/action_masking/test_agents_use_action_masks.py
@@ -100,6 +100,7 @@ def test_ray_single_agent_action_masking(monkeypatch):
     monkeypatch.undo()
 
 
+@pytest.mark.xfail(reason="Fails due to being flaky when run in CI.")
 def test_ray_multi_agent_action_masking(monkeypatch):
     """Check that Ray agents never take invalid actions when using MARL."""
     with open(MARL_PATH, "r") as f: