diff --git a/python/pyspark/streaming_tests.py b/python/pyspark/streaming_tests.py
index fe893210ab089..8396c4f960e81 100644
--- a/python/pyspark/streaming_tests.py
+++ b/python/pyspark/streaming_tests.py
@@ -43,7 +43,7 @@ def setUp(self):
 
     def tearDown(self):
         # Do not call pyspark.streaming.context.StreamingContext.stop directly because
-        # we do not wait to shutdown call back server and py4j client
+        # we do not wait to shutdown py4j client.
         self.ssc._jssc.stop()
         self.ssc._sc.stop()
         # Why does it long time to terminate StremaingContext and SparkContext?
@@ -74,7 +74,6 @@ def setUp(self):
         PySparkStreamingTestCase.setUp(self)
         self.timeout = 10  # seconds
         self.numInputPartitions = 2
-        self.result = list()
 
     def tearDown(self):
         PySparkStreamingTestCase.tearDown(self)
@@ -426,7 +425,8 @@ def _run_stream(self, test_input, test_func, expected_output, numSlices=None):
         # Apply test function to stream.
         test_stream = test_func(test_input_stream)
         # Add job to get output from stream.
-        test_stream._test_output(self.result)
+        result = list()
+        test_stream._test_output(result)
         self.ssc.start()
 
         start_time = time.time()
@@ -438,10 +438,10 @@ def _run_stream(self, test_input, test_func, expected_output, numSlices=None):
                 break
             self.ssc.awaitTermination(50)
             # Check if the output is the same length of expexted output.
-            if len(expected_output) == len(self.result):
+            if len(expected_output) == len(result):
                 break
 
-        return self.result
+        return result
 
 class TestSaveAsFilesSuite(PySparkStreamingTestCase):
     def setUp(self):