diff --git a/Lib/test/test_sgmllib.py b/Lib/test/test_sgmllib.py
new file mode 100644
index 0000000..a37696d
--- /dev/null
+++ b/Lib/test/test_sgmllib.py
@@ -0,0 +1,243 @@
+import pprint
+import sgmllib
+import test_support
+import unittest
+
+
+class EventCollector(sgmllib.SGMLParser):
+
+    def __init__(self):
+        self.events = []
+        self.append = self.events.append
+        sgmllib.SGMLParser.__init__(self)
+
+    def get_events(self):
+        # Normalize the list of events so that buffer artefacts don't
+        # separate runs of contiguous characters.
+        L = []
+        prevtype = None
+        for event in self.events:
+            type = event[0]
+            if type == prevtype == "data":
+                L[-1] = ("data", L[-1][1] + event[1])
+            else:
+                L.append(event)
+            prevtype = type
+        self.events = L
+        return L
+
+    # structure markup
+
+    def unknown_starttag(self, tag, attrs):
+        self.append(("starttag", tag, attrs))
+
+    def unknown_endtag(self, tag):
+        self.append(("endtag", tag))
+
+    # all other markup
+
+    def handle_comment(self, data):
+        self.append(("comment", data))
+
+    def handle_charref(self, data):
+        self.append(("charref", data))
+
+    def handle_data(self, data):
+        self.append(("data", data))
+
+    def handle_decl(self, decl):
+        self.append(("decl", decl))
+
+    def handle_entityref(self, data):
+        self.append(("entityref", data))
+
+    def handle_pi(self, data):
+        self.append(("pi", data))
+
+
+class CDATAEventCollector(EventCollector):
+    def start_cdata(self, attrs):
+        self.append(("starttag", "cdata", attrs))
+        self.setliteral()
+
+
+class SGMLParserTestCase(unittest.TestCase):
+
+    collector = EventCollector
+
+    def check_events(self, source, expected_events):
+        parser = self.collector()
+        for s in source:
+            parser.feed(s)
+        parser.close()
+        events = parser.get_events()
+        if events != expected_events:
+            self.fail("received events did not match expected events\n"
+                      "Expected:\n" + pprint.pformat(expected_events) +
+                      "\nReceived:\n" + pprint.pformat(events))
+
+    def check_parse_error(self, source):
+        parser = EventCollector()
+        try:
+            parser.feed(source)
+            parser.close()
+        except sgmllib.SGMLParseError:
+            pass
+        else:
+            self.fail("expected SGMLParseError for %r\nReceived:\n%s"
+                      % (source, pprint.pformat(parser.get_events())))
+
+    def test_underscore_in_attrname(self):
+        # SF bug #436621
+        """Make sure attribute names with underscores are accepted"""
+        self.check_events("<a has_under _under>", [
+            ("starttag", "a", [("has_under", "has_under"),
+                               ("_under", "_under")]),
+            ])
+
+    def test_underscore_in_tagname(self):
+        # SF bug #436621
+        """Make sure tag names with underscores are accepted"""
+        self.check_events("<has_under></has_under>", [
+            ("starttag", "has_under", []),
+            ("endtag", "has_under"),
+            ])
+
+    def test_quotes_in_unquoted_attrs(self):
+        # SF bug #436621
+        """Be sure quotes in unquoted attributes are made part of the value"""
+        self.check_events("<a href=foo'bar\"baz>", [
+            ("starttag", "a", [("href", "foo'bar\"baz")]),
+            ])
+
+    def test_xhtml_empty_tag(self):
+        """Handling of XHTML-style empty start tags"""
+        self.check_events("<br />text<i></i>", [
+            ("starttag", "br", []),
+            ("data", "text"),
+            ("starttag", "i", []),
+            ("endtag", "i"),
+            ])
+
+    def test_processing_instruction_only(self):
+        self.check_events("<?processing instruction>", [
+            ("pi", "processing instruction"),
+            ])
+
+    def test_bad_nesting(self):
+        self.check_events("<a><b></a></b>", [
+            ("starttag", "a", []),
+            ("starttag", "b", []),
+            ("endtag", "a"),
+            ("endtag", "b"),
+            ])
+
+    def test_attr_syntax(self):
+        output = [
+          ("starttag", "a", [("b", "v"), ("c", "v"), ("d", "v"), ("e", "e")])
+          ]
+        self.check_events("""<a b='v' c="v" d=v e>""", output)
+        self.check_events("""<a  b = 'v' c = "v" d = v e>""", output)
+        self.check_events("""<a\nb\n=\n'v'\nc\n=\n"v"\nd\n=\nv\ne>""", output)
+        self.check_events("""<a\tb\t=\t'v'\tc\t=\t"v"\td\t=\tv\te>""", output)
+
+    def test_attr_values(self):
+        self.check_events("""<a b='xxx\n\txxx' c="yyy\t\nyyy" d='\txyz\n'>""",
+                        [("starttag", "a", [("b", "xxx\n\txxx"),
+                                            ("c", "yyy\t\nyyy"),
+                                            ("d", "\txyz\n")])
+                         ])
+        self.check_events("""<a b='' c="">""", [
+            ("starttag", "a", [("b", ""), ("c", "")]),
+            ])
+
+    def test_attr_funky_names(self):
+        self.check_events("""<a a.b='v' c:d=v e-f=v>""", [
+            ("starttag", "a", [("a.b", "v"), ("c:d", "v"), ("e-f", "v")]),
+            ])
+
+    def test_weird_starttags(self):
+        self.check_events("<a<a>", [
+            ("starttag", "a", []),
+            ("starttag", "a", []),
+            ])
+        self.check_events("</a<a>", [
+            ("endtag", "a"),
+            ("starttag", "a", []),
+            ])
+
+    def test_declaration_junk_chars(self):
+        self.check_parse_error("<!DOCTYPE foo $ >")
+
+    def test_get_starttag_text(self):
+        s = """<foobar   \n   one="1"\ttwo=2   >"""
+        self.check_events(s, [
+            ("starttag", "foobar", [("one", "1"), ("two", "2")]),
+            ])
+
+    def test_cdata_content(self):
+        s = ("<cdata> <!-- not a comment --> &not-an-entity-ref; </cdata>"
+             "<notcdata> <!-- comment --> </notcdata>")
+        self.collector = CDATAEventCollector
+        self.check_events(s, [
+            ("starttag", "cdata", []),
+            ("data", " <!-- not a comment --> &not-an-entity-ref; "),
+            ("endtag", "cdata"),
+            ("starttag", "notcdata", []),
+            ("data", " "),
+            ("comment", " comment "),
+            ("data", " "),
+            ("endtag", "notcdata"),
+            ])
+        s = """<cdata> <not a='start tag'> </cdata>"""
+        self.check_events(s, [
+            ("starttag", "cdata", []),
+            ("data", " <not a='start tag'> "),
+            ("endtag", "cdata"),
+            ])
+
+    # XXX These tests have been disabled by prefixing their names with
+    # an underscore.  The first two exercise outstanding bugs in the
+    # sgmllib module, and the third exhibits questionable behavior
+    # that needs to be carefully considered before changing it.
+
+    def _test_starttag_end_boundary(self):
+        self.check_events("""<a b='<'>""", [("starttag", "a", [("b", "<")])])
+        self.check_events("""<a b='>'>""", [("starttag", "a", [("b", ">")])])
+
+    def _test_buffer_artefacts(self):
+        output = [("starttag", "a", [("b", "<")])]
+        self.check_events(["<a b='<'>"], output)
+        self.check_events(["<a ", "b='<'>"], output)
+        self.check_events(["<a b", "='<'>"], output)
+        self.check_events(["<a b=", "'<'>"], output)
+        self.check_events(["<a b='<", "'>"], output)
+        self.check_events(["<a b='<'", ">"], output)
+
+        output = [("starttag", "a", [("b", ">")])]
+        self.check_events(["<a b='>'>"], output)
+        self.check_events(["<a ", "b='>'>"], output)
+        self.check_events(["<a b", "='>'>"], output)
+        self.check_events(["<a b=", "'>'>"], output)
+        self.check_events(["<a b='>", "'>"], output)
+        self.check_events(["<a b='>'", ">"], output)
+
+    def _test_starttag_junk_chars(self):
+        self.check_parse_error("<")
+        self.check_parse_error("<>")
+        self.check_parse_error("</$>")
+        self.check_parse_error("</")
+        self.check_parse_error("</a")
+        self.check_parse_error("<$")
+        self.check_parse_error("<$>")
+        self.check_parse_error("<!")
+        self.check_parse_error("<a $>")
+        self.check_parse_error("<a")
+        self.check_parse_error("<a foo='bar'")
+        self.check_parse_error("<a foo='bar")
+        self.check_parse_error("<a foo='>'")
+        self.check_parse_error("<a foo='>")
+        self.check_parse_error("<a foo=>")
+
+
+test_support.run_unittest(SGMLParserTestCase)