diff --git a/scrapy/tests/test_utils_url.py b/scrapy/tests/test_utils_url.py
index ec27bddea..d99ce9103 100644
--- a/scrapy/tests/test_utils_url.py
+++ b/scrapy/tests/test_utils_url.py
@@ -60,12 +60,15 @@ class UrlUtilsTest(unittest.TestCase):
 
     def test_canonicalize_url(self):
         # simplest case
-        self.assertEqual(canonicalize_url("http://www.example.com"),
-                                          "http://www.example.com")
+        self.assertEqual(canonicalize_url("http://www.example.com/"),
+                                          "http://www.example.com/")
 
         # always return a str
         assert isinstance(canonicalize_url(u"http://www.example.com"), str)
 
+        # append missing path
+        self.assertEqual(canonicalize_url("http://www.example.com"),
+                                          "http://www.example.com/")
         # typical usage
         self.assertEqual(canonicalize_url("http://www.example.com/do?a=1&b=2&c=3"),
                                           "http://www.example.com/do?a=1&b=2&c=3")
@@ -145,8 +148,8 @@ class UrlUtilsTest(unittest.TestCase):
                                            'http://www.example.com/caf%E9-con-leche.htm')
 
         # domains are case insensitive
-        self.assertEqual(canonicalize_url("http://www.EXAMPLE.com"),
-                                          "http://www.example.com")
+        self.assertEqual(canonicalize_url("http://www.EXAMPLE.com/"),
+                                          "http://www.example.com/")
 
 
 if __name__ == "__main__":
diff --git a/scrapy/utils/url.py b/scrapy/utils/url.py
index cae73f398..4e834681b 100644
--- a/scrapy/utils/url.py
+++ b/scrapy/utils/url.py
@@ -52,7 +52,7 @@ def canonicalize_url(url, keep_blank_values=True, keep_fragments=False, \
     keyvals = cgi.parse_qsl(query, keep_blank_values)
     keyvals.sort()
     query = urllib.urlencode(keyvals)
-    path = safe_url_string(urllib.unquote(path))
+    path = safe_url_string(urllib.unquote(path)) or '/'
     fragment = '' if not keep_fragments else fragment
     return urlparse.urlunparse((scheme, netloc.lower(), path, params, query, fragment))