diff --git a/awscli/formatter.py b/awscli/formatter.py
index e13b6fd72c62..a4cd3a789069 100644
--- a/awscli/formatter.py
+++ b/awscli/formatter.py
@@ -231,6 +231,20 @@ def __call__(self, command_name, response, stream=None):
         if stream is None:
             stream = self._get_default_stream()
         try:
+            # For paginated responses + a JMESPath query, fully buffer the
+            # result before applying the query.
+            #
+            # Streaming per-page + applying the query per-page can yield
+            # confusing extra output when later pages do not contain the
+            # queried field (for example: a scalar query printing a trailing
+            # "None" line). Buffering keeps text output consistent with the
+            # JSON/table formatters.
+            if is_response_paginated(response) and self._args.query is not None:
+                response_data = response.build_full_result()
+                self._remove_request_id(response_data)
+                self._format_response(response_data, stream)
+                return
+
             if is_response_paginated(response):
                 result_keys = response.result_keys
                 for i, page in enumerate(response):
@@ -246,8 +260,6 @@ def __call__(self, command_name, response, stream=None):
                         )
                     self._format_response(current, stream)
                 if response.resume_token:
-                    # Tell the user about the next token so they can continue
-                    # if they want.
                     self._format_response(
                         {'NextToken': {'NextToken': response.resume_token}},
                         stream,
@@ -256,8 +268,6 @@ def __call__(self, command_name, response, stream=None):
                 self._remove_request_id(response)
                 self._format_response(response, stream)
         finally:
-            # flush is needed to avoid the "close failed in file object
-            # destructor" in python2.x (see http://bugs.python.org/issue11380).
             self._flush_stream(stream)
 
     def _format_response(self, response, stream):
diff --git a/tests/unit/test_text_formatter_pagination.py b/tests/unit/test_text_formatter_pagination.py
new file mode 100644
index 000000000000..efcd6fda4223
--- /dev/null
+++ b/tests/unit/test_text_formatter_pagination.py
@@ -0,0 +1,44 @@
+import io
+
+import jmespath
+from botocore.paginate import PageIterator
+
+from awscli.formatter import TextFormatter
+
+
+class FakePageIterator(PageIterator):
+    # We intentionally don't call the parent constructor; this object is only
+    # used to satisfy `isinstance(response, PageIterator)` and provide a stable
+    # `build_full_result()` for the formatter.
+    def __init__(self, full_result):
+        self._full_result = full_result
+
+    def build_full_result(self):
+        return self._full_result
+
+
+class Args:
+    def __init__(self, query):
+        self.query = query
+
+
+def test_text_formatter_buffers_paginated_response_when_query_is_set():
+    # Regression for aws/aws-cli#10061:
+    # when a response is paginated, text output previously applied the query
+    # per-page. If later pages didn't include the queried scalar field, the
+    # formatter could emit an extra trailing "None" line.
+    args = Args(query=jmespath.compile('ChangeSetId'))
+    formatter = TextFormatter(args)
+
+    response = FakePageIterator(
+        {
+            'ResponseMetadata': {'RequestId': 'abc'},
+            'ChangeSetId': 'arn:aws:cloudformation:us-east-1:000000000000:changeSet/change-set-1/11111111111111111111111111',
+            'NextToken': None,
+        }
+    )
+
+    stream = io.StringIO()
+    formatter('cloudformation describe-change-set', response, stream=stream)
+
+    assert stream.getvalue().strip() == response._full_result['ChangeSetId']