diff --git a/CHANGES.txt b/CHANGES.txt
index c7f32ea4..a1b2dae3 100644
--- a/CHANGES.txt
+++ b/CHANGES.txt
@@ -4,6 +4,11 @@ Unreleased
 Bugfix
 ~~~~~~
 
+- Waitress now normalizes a non-ASCII ``url_prefix`` to the PEP 3333
+  encoding used for request paths, so the prefix matches the parsed
+  ``PATH_INFO``/``SCRIPT_NAME``. See
+  https://github.com/Pylons/waitress/pull/493.
+
 - Renamed the HTTP header "Trailers" to "Trailer" to fix a typo and comply with
   the correct header name as specified in RFC 7230.
 
diff --git a/src/waitress/adjustments.py b/src/waitress/adjustments.py
index 8b26eb87..9389c4ed 100644
--- a/src/waitress/adjustments.py
+++ b/src/waitress/adjustments.py
@@ -75,6 +75,11 @@ def slash_fixed_str(s):
         # always have a leading slash, replace any number of leading slashes
         # with a single slash, and strip any trailing slashes
         s = "/" + s.lstrip("/").rstrip("/")
+        # Normalize encoding to match PEP 3333: WSGI PATH_INFO/SCRIPT_NAME
+        # are raw URL bytes decoded as ISO-8859-1.  Encode the user-supplied
+        # Unicode string as UTF-8 and decode as ISO-8859-1 so that non-ASCII
+        # characters compare equal to the parser-produced path.
+        s = s.encode("utf-8").decode("iso-8859-1")
     return s
 
 
diff --git a/tests/test_adjustments.py b/tests/test_adjustments.py
index 86bf5ded..591b3ac4 100644
--- a/tests/test_adjustments.py
+++ b/tests/test_adjustments.py
@@ -384,6 +384,27 @@ def test_server_header_removable(self):
         inst = self._makeOne(ident="specific_header")
         self.assertEqual(inst.ident, "specific_header")
 
+    def test_url_prefix_unicode_normalization(self):
+        from waitress.parser import unquote_bytes_to_wsgi
+
+        # Non-ASCII url_prefix should be normalized to PEP 3333 format
+        # (UTF-8-encoded bytes decoded as ISO-8859-1) so it compares
+        # equal to the parser-produced path for the same logical path.
+        inst = self._makeOne(url_prefix="/\xfc/")  # ü as latin-1 byte
+        # \xfc (ü in latin-1) encoded as UTF-8 is \xc3\xbc.
+        # Decoded as latin-1 that gives two code points: \xc3 \xbc.
+        # Note: slash_fixed_str strips any trailing slash.
+        self.assertEqual(inst.url_prefix, "/ü")
+
+        # ASCII-only url_prefix must still be unchanged.
+        inst2 = self._makeOne(url_prefix="/api")
+        self.assertEqual(inst2.url_prefix, "/api")
+
+        # Now check it matches a parser-produced path for /ü
+        raw_uri = b"/%C3%BC"
+        path = unquote_bytes_to_wsgi(raw_uri)
+        self.assertEqual(inst.url_prefix, path)
+
 
 class TestCLI(unittest.TestCase):
     def parse(self, argv):