Raise error in read_csv when arguments header and prefix both are not None

doc/source/whatsnew/v1.1.0.rst

@@ -162,7 +162,7 @@ MultiIndex
 I/O
 ^^^
 - Bug in :meth:`read_json` where integer overflow was occuring when json contains big number strings. (:issue:`30320`)
--
+- `read_csv` will now raise a ``ValueError`` when arguments `header` and `prefix` both are not None. (:issue:`27394`)
 -
 
 Plotting

pandas/io/parsers.py

@@ -1888,6 +1888,10 @@ def __init__(self, src, **kwds):
         if self._reader.header is None:
             self.names = None
         else:
+            if self.prefix:
+                raise ValueError(
+                    "Argument prefix must be None if argument header is not None"
+                )
             if len(self._reader.header) > 1:
                 # we have a multi index in the columns
                 (

pandas/tests/frame/test_to_csv.py

@@ -575,6 +575,13 @@ def test_to_csv_headers(self):
             recons.reset_index(inplace=True)
             tm.assert_frame_equal(to_df, recons)
 
+    def test_to_csv_raises_on_header_prefix(self):
+        # gh-27394
+        msg = "Argument prefix must be None if argument header is not None"
+        s = StringIO("0,1\n2,3")
+        with pytest.raises(ValueError, match=msg):
+            read_csv(s, header=0, prefix="_X")
+
     def test_to_csv_multiindex(self, float_frame, datetime_frame):
 
         frame = float_frame