diff --git a/test_framework/runner.py b/test_framework/runner.py
index d690f790..df86239b 100644
--- a/test_framework/runner.py
+++ b/test_framework/runner.py
@@ -241,7 +241,11 @@ def parse_arguments() -> argparse.Namespace:
     # if --check-setup is present, shouldn't have any other options
     if args.check_setup:
         ignored_args = [
-            k for k, v in vars(args).items() if bool(v) and (k != "check_setup")
+            k
+            for k, v in vars(args).items()
+            if bool(v)
+            and (k != "check_setup")
+            and (k != "stage" or (k == "stage" and v != "run"))
         ]
         if ignored_args:
             warnings.warn(
@@ -273,7 +277,6 @@ def parse_arguments() -> argparse.Namespace:
         warnings.warn("Option --no-coalescing has no impact on Part I & Part II tests")
 
     if args.expected_error_codes:
-
         out_of_range = [str(i) for i in args.expected_error_codes if i < 1 or i > 255]
         if out_of_range:
             bad_codes = ", ".join(out_of_range)
@@ -436,7 +439,10 @@ def main() -> int:
     """Main entry point for test runner"""
     args = parse_arguments()
     if args.check_setup:
-        return check_setup()
+        success = check_setup()
+        if success:
+            return 0
+        return 1
 
     compiler = Path(args.cc).resolve()
 
diff --git a/test_framework/test_tests/test_toplevel.py b/test_framework/test_tests/test_toplevel.py
index d15c300f..09849833 100644
--- a/test_framework/test_tests/test_toplevel.py
+++ b/test_framework/test_tests/test_toplevel.py
@@ -91,6 +91,27 @@ def get_failure_count(failure: subprocess.CalledProcessError) -> int:
 
 
 class TopLevelTest(unittest.TestCase):
+    def test_check_setup(self) -> None:
+        """Test the check-setup command"""
+        try:
+            check = run_test_script("./test_compiler --check-setup")
+            self.assertEqual(check.stdout, "All system requirements met!\n")
+            self.assertEqual(check.stderr, "")
+        except subprocess.CalledProcessError as err:
+            # Test should still pass if only problem is that GDB isn't installed (it isn't on Github's runners)
+            if (
+                err.stdout.startswith(
+                    "No debugger found. The test script doesn't require a debugger but you probably want one for, ya know, debugging."
+                )
+                and err.stderr == ""
+            ):
+                # okay
+                return
+            # we found some other error
+            self.fail(
+                f"--check-setup option failed.\nstderr:\n{err.stderr}\nstdout:\n{err.stdout}"
+            )
+
     def test_one_chapter(self) -> None:
         """We can run tests for a single chapter with --latest-only"""
         expected_test_count = get_expected_test_count(chapters=[2])