diff --git a/cssselect/parser.py b/cssselect/parser.py
index 898b2bc..1560823 100644
--- a/cssselect/parser.py
+++ b/cssselect/parser.py
@@ -997,7 +997,7 @@ def _compile(pattern: str) -> MatchFunc:
 
 def _replace_unicode(match: re.Match[str]) -> str:
     codepoint = int(match.group(1), 16)
-    if codepoint > sys.maxunicode or 0xD800 <= codepoint <= 0xDFFF:
+    if codepoint == 0 or codepoint > sys.maxunicode or 0xD800 <= codepoint <= 0xDFFF:
         codepoint = 0xFFFD
     return chr(codepoint)
 
diff --git a/tests/test_cssselect.py b/tests/test_cssselect.py
index 5b98711..661f663 100644
--- a/tests/test_cssselect.py
+++ b/tests/test_cssselect.py
@@ -1076,6 +1076,11 @@ def test_unicode_escapes(self) -> None:
         )
         # A code point beyond the Unicode range is replaced with U+FFFD.
         assert css_to_xpath(r"\110000") == ("descendant-or-self::*[name() = '\ufffd']")
+        # A null code point is likewise replaced with U+FFFD, so no NUL
+        # character can leak into the generated XPath string.
+        assert css_to_xpath(r"*[aval='\0 ']") == (
+            "descendant-or-self::*[@aval = '\ufffd']"
+        )
         # unescape_ident() resolves both unicode and simple escapes.
         assert unescape_ident(r"\41 B") == "AB"
         assert unescape_ident(r"\-foo") == "-foo"