diff --git a/rdflib/plugins/parsers/ntriples.py b/rdflib/plugins/parsers/ntriples.py
index 99b966b1b..67dbe9d7b 100644
--- a/rdflib/plugins/parsers/ntriples.py
+++ b/rdflib/plugins/parsers/ntriples.py
@@ -28,7 +28,7 @@
 
 __all__ = ['unquote', 'uriquote', 'Sink', 'NTriplesParser']
 
-uriref = r'<([^:]+:[^\s"<>]+)>'
+uriref = r'<([^:]+:[^\s"<>]*)>'
 literal = r'"([^"\\]*(?:\\.[^"\\]*)*)"'
 litinfo = r'(?:@([a-zA-Z]+(?:-[a-zA-Z0-9]+)*)|\^\^' + uriref + r')?'
 
diff --git a/test/test_issue920.py b/test/test_issue920.py
new file mode 100644
index 000000000..eb12edc49
--- /dev/null
+++ b/test/test_issue920.py
@@ -0,0 +1,36 @@
+"""
+Issue 920 - NTriples fails to parse URIs with only a scheme
+
+from rdflib import Graph
+g=Graph()
+g.parse(data='<a:> <b:> <c:> .', format='nt') # nquads also fails
+
+N3, by contrast, succeeds:
+
+g.parse(data='<a:> <b:> <c:> .', format='n3')
+"""
+from rdflib import Graph
+import unittest
+
+
+class TestIssue920(unittest.TestCase):
+
+    def test_issue_920(self):
+        g = Graph()
+        # NT tests
+        g.parse(data='<a:> <b:> <c:> .', format='nt')
+        g.parse(data='<http://a> <http://b> <http://c> .', format='nt')
+        g.parse(data='<https://a> <http://> <http://c> .', format='nt')
+
+        # related parser tests
+        g.parse(data='<a:> <b:> <c:> .', format='turtle')
+        g.parse(data='<http://a> <http://b> <http://c> .', format='turtle')
+        g.parse(data='<https://a> <http://> <http://c> .', format='turtle')
+
+        g.parse(data='<a:> <b:> <c:> .', format='n3')
+        g.parse(data='<http://a> <http://b> <http://c> .', format='n3')
+        g.parse(data='<https://a> <http://> <http://c> .', format='n3')
+
+
+if __name__ == "__main__":
+    unittest.main()