diff --git a/src/python/turicreate/data_structures/sframe.py b/src/python/turicreate/data_structures/sframe.py
index cb5db0c05f..b8206b07a0 100644
--- a/src/python/turicreate/data_structures/sframe.py
+++ b/src/python/turicreate/data_structures/sframe.py
@@ -1603,13 +1603,18 @@ def read_json(cls,
         [3 rows x 1 columns]
         """
         if orient == "records":
-            g = SArray.read_json(url)
-            if len(g) == 0:
-                return SFrame()
-            if g.dtype != dict:
-                raise RuntimeError("Invalid input JSON format. Expected list of dictionaries")
-            g = SFrame({'X1':g})
-            return g.unpack('X1','')
+            if type(url)==str and url[0] !='{' :
+                g = SArray.read_json(url)
+                if len(g) == 0:
+                    return SFrame()
+                if g.dtype != dict:
+                    raise RuntimeError("Invalid input JSON format. Expected list of dictionaries")
+                g = SFrame({'X1':g})
+                return g.unpack('X1','')
+            elif type(url)==str and url[0] =='{' :
+                url=pandas.read_json(url)
+                url=SFrame(url)
+                return url
         elif orient == "lines":
             g = cls.read_csv(url, header=False,na_values=['null'],true_values=['true'],false_values=['false'],
                     _only_raw_string_substitutions=True)
diff --git a/src/python/turicreate/test/test_sframe.py b/src/python/turicreate/test/test_sframe.py
index 91a0c8cfcd..a7e27bfa21 100644
--- a/src/python/turicreate/test/test_sframe.py
+++ b/src/python/turicreate/test/test_sframe.py
@@ -201,6 +201,13 @@ def test_auto_parse_csv_with_bom(self):
             sf = SFrame.read_csv(csvfile.name, header=True)
             self.assertEqual(sf.dtype, [float, int, str])
             self.__test_equal(sf, df)
+    def test_read_json(self):
+        x='{"row 1":{"col 1":"a","col 2":"b"},"row 2":{"col 1":"c","col 2":"d"}}'
+        sf=SFrame.read_json(x)
+        df =pd.read_json(x)
+        df=df.reset_index()
+        df=df.drop(['index'],axis=1)
+        self.__test_equal(sf, df)
 
     def test_auto_parse_csv(self):
         with tempfile.NamedTemporaryFile(mode='w', delete=False) as csvfile: