to_csv headers kwarg now works regardless of index kwarg

doc/source/release.rst

@@ -381,6 +381,8 @@ Bug Fixes
 - Bug in arithmetic operations affecting to NaT (:issue:`6873`)
 - Bug in ``Series.str.extract`` where the resulting ``Series`` from a single
   group match wasn't renamed to the group name
+- Bug in ``DataFrame.to_csv`` where setting `index` to `False` ignored the
+  `header` kwarg (:issue:`6186`)
 
 pandas 0.13.1
 -------------

pandas/core/format.py

@@ -1194,6 +1194,14 @@ def _save_header(self):
         has_aliases = isinstance(header, (tuple, list, np.ndarray))
         if not (has_aliases or self.header):
             return
+        if has_aliases:
+            if len(header) != len(cols):
+                raise ValueError(('Writing %d cols but got %d aliases'
+                                  % (len(cols), len(header))))
+            else:
+                write_cols = header
+        else:
+            write_cols = cols
 
         if self.index:
             # should write something for index label
@@ -1219,22 +1227,8 @@ def _save_header(self):
             else:
                 encoded_labels = []
 
-            if has_aliases:
-                if len(header) != len(cols):
-                    raise ValueError(('Writing %d cols but got %d aliases'
-                                      % (len(cols), len(header))))
-                else:
-                    write_cols = header
-            else:
-                write_cols = cols
-
-            if not has_mi_columns:
-                encoded_labels += list(write_cols)
-
-        else:
-
-            if not has_mi_columns:
-                encoded_labels += list(cols)
+        if not has_mi_columns:
+            encoded_labels += list(write_cols)
 
         # write out the mi
         if has_mi_columns:

pandas/tests/test_frame.py

@@ -5836,6 +5836,22 @@ def test_to_csv_no_index(self):
             result = read_csv(path)
             assert_frame_equal(df,result)
 
+    def test_to_csv_headers(self):
+        # GH6186, the presence or absence of `index` incorrectly
+        # causes to_csv to have different header semantics.
+        pname = '__tmp_to_csv_headers__'
+        from_df = DataFrame([[1, 2], [3, 4]], columns=['A', 'B'])
+        to_df  = DataFrame([[1, 2], [3, 4]], columns=['X', 'Y'])
+        with ensure_clean(pname) as path:
+            from_df.to_csv(path, header=['X', 'Y'])
+            recons = DataFrame.from_csv(path)
+            assert_frame_equal(to_df, recons)
+
+            from_df.to_csv(path, index=False, header=['X', 'Y'])
+            recons = DataFrame.from_csv(path)
+            recons.reset_index(inplace=True)
+            assert_frame_equal(to_df, recons)
+
     def test_to_csv_multiindex(self):
 
         pname = '__tmp_to_csv_multiindex__'