diff --git a/tensorflow/python/kernel_tests/cast_op_test.py b/tensorflow/python/kernel_tests/cast_op_test.py
index f26a1b6b58d..e9be8e7d5f7 100644
--- a/tensorflow/python/kernel_tests/cast_op_test.py
+++ b/tensorflow/python/kernel_tests/cast_op_test.py
@@ -213,13 +213,6 @@ class CastOpTest(test.TestCase):
           err = gradient_checker.compute_gradient_error(x, [], y, [])
           self.assertLess(err, 1e-3)
 
-  def testPythonDataTypes(self):
-    with self.cached_session():
-      # GitHub issue 35938, a of 0.2 is for python native type.
-      a = 0.2
-      b = math_ops.cast(a, dtypes.float64)
-      self.assertAllEqual(a, self.evaluate(b))
-
 
 class SparseTensorCastTest(test.TestCase):
 
diff --git a/tensorflow/python/ops/math_ops.py b/tensorflow/python/ops/math_ops.py
index b717077a80d..ee7b1194610 100644
--- a/tensorflow/python/ops/math_ops.py
+++ b/tensorflow/python/ops/math_ops.py
@@ -745,7 +745,7 @@ def cast(x, dtype, name=None):
       # ops.convert_to_tensor(x, dtype=dtype, ...)  here, but that
       # allows some conversions that cast() can't do, e.g. casting numbers to
       # strings.
-      x = ops.convert_to_tensor(x, dtype_hint=base_type, name="x")
+      x = ops.convert_to_tensor(x, name="x")
       if x.dtype.base_dtype != base_type:
         x = gen_math_ops.cast(x, base_type, name=name)
     if x.dtype.is_complex and base_type.is_floating: