diff --git a/Lib/encodings/punycode.py b/Lib/encodings/punycode.py
--- a/Lib/encodings/punycode.py
+++ b/Lib/encodings/punycode.py
@@ -180,17 +180,19 @@ def insertion_sort(base, extended, error
     return base
 
 def punycode_decode(text, errors):
+    if errors not in ('strict', 'ignore'):
+        raise UnicodeError("Unsupported error handling %s" % errors)
     if isinstance(text, str):
-        text = text.encode("ascii")
+        text = text.encode("ascii", errors)
     if isinstance(text, memoryview):
         text = bytes(text)
     pos = text.rfind(b"-")
     if pos == -1:
         base = ""
-        extended = str(text, "ascii").upper()
+        extended = str(text, "ascii", errors).upper()
     else:
         base = str(text[:pos], "ascii", errors)
-        extended = str(text[pos+1:], "ascii").upper()
+        extended = str(text[pos+1:], "ascii", errors).upper()
     return insertion_sort(base, extended, errors)
 
 ### Codec APIs
@@ -202,8 +204,6 @@ class Codec(codecs.Codec):
         return res, len(input)
 
     def decode(self, input, errors='strict'):
-        if errors not in ('strict', 'replace', 'ignore'):
-            raise UnicodeError("Unsupported error handling "+errors)
         res = punycode_decode(input, errors)
         return res, len(input)
 
@@ -213,8 +213,6 @@ class IncrementalEncoder(codecs.Incremen
 
 class IncrementalDecoder(codecs.IncrementalDecoder):
     def decode(self, input, final=False):
-        if self.errors not in ('strict', 'replace', 'ignore'):
-            raise UnicodeError("Unsupported error handling "+self.errors)
         return punycode_decode(input, self.errors)
 
 class StreamWriter(Codec,codecs.StreamWriter):
diff --git a/Lib/test/test_codecs.py b/Lib/test/test_codecs.py
--- a/Lib/test/test_codecs.py
+++ b/Lib/test/test_codecs.py
@@ -882,6 +882,10 @@ class PunycodeTest(unittest.TestCase):
             puny = puny.decode("ascii").encode("ascii")
             self.assertEqual(uni, puny.decode("punycode"))
 
+    def test_decode_errors(self):
+        self.assertEqual(b'a\xff-\xffb\xffga\xff'.decode("punycode", "ignore"), "a\xe9")
+        self.assertRaises(UnicodeError, b'abc-'.decode, "punycode", "replace")
+
 class UnicodeInternalTest(unittest.TestCase):
     def test_bug1251300(self):
         # Decoding with unicode_internal used to not correctly handle "code