diff --git a/changelog/5547.bugfix.rst b/changelog/5547.bugfix.rst
new file mode 100644
index 00000000000..0345bf048fc
--- /dev/null
+++ b/changelog/5547.bugfix.rst
@@ -0,0 +1 @@
+``--step-wise`` now handles ``xfail(strict=True)`` markers properly.
diff --git a/src/_pytest/stepwise.py b/src/_pytest/stepwise.py
index eb45554904a..a18a58573bd 100644
--- a/src/_pytest/stepwise.py
+++ b/src/_pytest/stepwise.py
@@ -72,7 +72,7 @@ def pytest_collection_modifyitems(self, session, config, items):
 
     def pytest_runtest_logreport(self, report):
         # Skip this hook if plugin is not active or the test is xfailed.
-        if not self.active or "xfail" in report.keywords:
+        if not self.active:
             return
 
         if report.failed:
diff --git a/testing/test_stepwise.py b/testing/test_stepwise.py
index 40c86fec3d8..591d67b6ce7 100644
--- a/testing/test_stepwise.py
+++ b/testing/test_stepwise.py
@@ -165,3 +165,56 @@ def test_stop_on_collection_errors(broken_testdir, broken_first):
         files.reverse()
     result = broken_testdir.runpytest("-v", "--strict-markers", "--stepwise", *files)
     result.stdout.fnmatch_lines("*errors during collection*")
+
+
+def test_xfail_handling(testdir):
+    """Ensure normal xfail is ignored, and strict xfail interrupts the session in sw mode
+
+    (#5547)
+    """
+    contents = """
+        import pytest
+        def test_a(): pass
+
+        @pytest.mark.xfail(strict={strict})
+        def test_b(): assert {assert_value}
+
+        def test_c(): pass
+        def test_d(): pass
+    """
+    testdir.makepyfile(contents.format(assert_value="0", strict="False"))
+    result = testdir.runpytest("--sw", "-v")
+    result.stdout.fnmatch_lines(
+        [
+            "*::test_a PASSED *",
+            "*::test_b XFAIL *",
+            "*::test_c PASSED *",
+            "*::test_d PASSED *",
+            "* 3 passed, 1 xfailed in *",
+        ]
+    )
+
+    testdir.makepyfile(contents.format(assert_value="1", strict="True"))
+    result = testdir.runpytest("--sw", "-v")
+    result.stdout.fnmatch_lines(
+        [
+            "*::test_a PASSED *",
+            "*::test_b FAILED *",
+            "* Interrupted*",
+            "* 1 failed, 1 passed in *",
+        ]
+    )
+
+    # because we are writing to the same file, mtime might not be affected enough to
+    # invalidate the cache, making this next run flaky
+    testdir.tmpdir.join("__pycache__").remove()
+    testdir.makepyfile(contents.format(assert_value="0", strict="True"))
+    result = testdir.runpytest("--sw", "-v")
+    result.stdout.fnmatch_lines(
+        [
+            "*::test_b XFAIL *",
+            "*::test_c PASSED *",
+            "*::test_d PASSED *",
+            "* 2 passed, 1 deselected, 1 xfailed in *",
+        ]
+    )