diff --git a/moto/s3/responses.py b/moto/s3/responses.py
index 8e56fb792..bd0dbc881 100644
--- a/moto/s3/responses.py
+++ b/moto/s3/responses.py
@@ -1,9 +1,12 @@
 from __future__ import unicode_literals
-import six
-from six.moves.urllib.parse import parse_qs, urlparse
+
 import re
 
+from boto.s3.key import Key
 from jinja2 import Template
+import six
+from six.moves.urllib.parse import parse_qs, urlparse
+
 
 from .exceptions import BucketAlreadyExists, MissingBucket
 from .models import s3_backend
@@ -232,8 +235,14 @@ class ResponseObject(object):
         for header, value in request.headers.items():
             if isinstance(header, six.string_types):
                 result = meta_regex.match(header)
+                meta_key = None
                 if result:
+                    # Check for extra metadata
                     meta_key = result.group(0).lower()
+                elif header.lower() in Key.base_user_settable_fields:
+                    # Check for special metadata that doesn't start with x-amz-meta
+                    meta_key = header
+                if meta_key:
                     metadata = request.headers[header]
                     key.set_metadata(meta_key, metadata)
 
diff --git a/tests/test_s3/test_s3.py b/tests/test_s3/test_s3.py
index 9a7561865..4f09c406f 100644
--- a/tests/test_s3/test_s3.py
+++ b/tests/test_s3/test_s3.py
@@ -641,3 +641,16 @@ def test_unicode_value():
     list(bucket.list())
     key = bucket.get_key(key.key)
     assert key.get_contents_as_string().decode("utf-8") == u'こんにちは.jpg'
+
+
+@mock_s3
+def test_setting_content_encoding():
+    conn = boto.connect_s3()
+    bucket = conn.create_bucket('mybucket')
+    key = bucket.new_key("keyname")
+    key.set_metadata("Content-Encoding", "gzip")
+    compressed_data = "abcdef"
+    key.set_contents_from_string(compressed_data)
+
+    key = bucket.get_key("keyname")
+    key.content_encoding.should.equal("gzip")