diff --git a/nova/rc_fields.py b/nova/rc_fields.py
index 8481089a7154..add6c76a60e2 100644
--- a/nova/rc_fields.py
+++ b/nova/rc_fields.py
@@ -58,11 +58,11 @@ class ResourceClass(fields.StringField):
     def normalize_name(cls, rc_name):
         if rc_name is None:
             return None
-        norm_name = rc_name.upper()
-        cust_prefix = cls.CUSTOM_NAMESPACE
-        norm_name = cust_prefix + norm_name
-        # Replace some punctuation characters with underscores
-        norm_name = re.sub('[^0-9A-Z]+', '_', norm_name)
+        # Replace non-alphanumeric characters with underscores
+        norm_name = re.sub('[^0-9A-Za-z]+', '_', rc_name)
+        # Bug #1762789: Do .upper after replacing non alphanumerics.
+        norm_name = norm_name.upper()
+        norm_name = cls.CUSTOM_NAMESPACE + norm_name
         return norm_name
 
 
diff --git a/nova/tests/unit/objects/test_fields.py b/nova/tests/unit/objects/test_fields.py
index 7868ab81c5bd..289b625ddcf4 100644
--- a/nova/tests/unit/objects/test_fields.py
+++ b/nova/tests/unit/objects/test_fields.py
@@ -370,13 +370,8 @@ class TestResourceClass(TestString):
         py3.  Make sure normalize_name handles it properly.
         """
         name = u'Fu\xdfball'
-        if six.PY2:
-            self.assertEqual(u'CUSTOM_FU_BALL',
-                             rc_fields.ResourceClass.normalize_name(name))
-        else:
-            # TODO(efried): When bug #1762789 is resolved, remove this branch.
-            self.assertEqual(u'CUSTOM_FUSSBALL',
-                             rc_fields.ResourceClass.normalize_name(name))
+        self.assertEqual(u'CUSTOM_FU_BALL',
+                         rc_fields.ResourceClass.normalize_name(name))
 
 
 class TestInteger(TestField):