diff --git a/python/pyspark/mllib/linalg/__init__.py b/python/pyspark/mllib/linalg/__init__.py
index ea42127f1651..d903b9030d8c 100644
--- a/python/pyspark/mllib/linalg/__init__.py
+++ b/python/pyspark/mllib/linalg/__init__.py
@@ -770,6 +770,9 @@ def __getitem__(self, index):
             raise ValueError("Index %d out of bounds." % index)
 
         insert_index = np.searchsorted(inds, index)
+        if insert_index >= inds.size:
+            return 0.
+
         row_ind = inds[insert_index]
         if row_ind == index:
             return vals[insert_index]
diff --git a/python/pyspark/mllib/tests.py b/python/pyspark/mllib/tests.py
index 96cf13495aa9..2a6a5cd3fe40 100644
--- a/python/pyspark/mllib/tests.py
+++ b/python/pyspark/mllib/tests.py
@@ -237,15 +237,17 @@ def test_conversion(self):
         self.assertTrue(dv.array.dtype == 'float64')
 
     def test_sparse_vector_indexing(self):
-        sv = SparseVector(4, {1: 1, 3: 2})
+        sv = SparseVector(5, {1: 1, 3: 2})
         self.assertEqual(sv[0], 0.)
         self.assertEqual(sv[3], 2.)
         self.assertEqual(sv[1], 1.)
         self.assertEqual(sv[2], 0.)
-        self.assertEqual(sv[-1], 2)
-        self.assertEqual(sv[-2], 0)
-        self.assertEqual(sv[-4], 0)
-        for ind in [4, -5]:
+        self.assertEqual(sv[4], 0.)
+        self.assertEqual(sv[-1], 0.)
+        self.assertEqual(sv[-2], 2.)
+        self.assertEqual(sv[-3], 0.)
+        self.assertEqual(sv[-5], 0.)
+        for ind in [5, -6]:
             self.assertRaises(ValueError, sv.__getitem__, ind)
         for ind in [7.8, '1']:
             self.assertRaises(TypeError, sv.__getitem__, ind)