diff --git a/kafka/producer/kafka.py b/kafka/producer/kafka.py
index 3ac578f18..291efccdd 100644
--- a/kafka/producer/kafka.py
+++ b/kafka/producer/kafka.py
@@ -664,7 +664,9 @@ def max_usable_produce_magic(cls, api_version):
         else:
             return 0
 
-    def _estimate_size_in_bytes(self, key, value, headers=[]):
+    def _estimate_size_in_bytes(self, key, value, headers=None):
+        if headers is None:
+            headers = []
         magic = self.max_usable_produce_magic(self.config['api_version'])
         if magic == 2:
             return DefaultRecordBatchBuilder.estimate_size_in_bytes(
diff --git a/kafka/record/memory_records.py b/kafka/record/memory_records.py
index dae2be732..79ddca737 100644
--- a/kafka/record/memory_records.py
+++ b/kafka/record/memory_records.py
@@ -162,11 +162,13 @@ def skip(self, offsets_to_skip):
         # Exposed for testing compacted records
         self._next_offset += offsets_to_skip
 
-    def append(self, timestamp, key, value, headers=[]):
+    def append(self, timestamp, key, value, headers=None):
         """ Append a message to the buffer.
 
         Returns: RecordMetadata or None if unable to append
         """
+        if headers is None:
+            headers = []
         if self._closed:
             return None