diff --git a/doc/source/whatsnew/v0.21.0.txt b/doc/source/whatsnew/v0.21.0.txt
index 3df0a21facb02..861bde7161519 100644
--- a/doc/source/whatsnew/v0.21.0.txt
+++ b/doc/source/whatsnew/v0.21.0.txt
@@ -83,6 +83,7 @@ Indexing
 I/O
 ^^^
 
+- Bug in ``pd.read_csv()`` in which non integer values for the header argument generated an unhelpful / unrelated error message (:issue:`16338`)
 
 
 Plotting
diff --git a/pandas/io/parsers.py b/pandas/io/parsers.py
index ce8643504932f..e287d92f67ef6 100755
--- a/pandas/io/parsers.py
+++ b/pandas/io/parsers.py
@@ -1164,6 +1164,8 @@ def __init__(self, kwds):
         # validate header options for mi
         self.header = kwds.get('header')
         if isinstance(self.header, (list, tuple, np.ndarray)):
+            if not all(map(is_integer, self.header)):
+                raise ValueError("header must be integer or list of integers")
             if kwds.get('as_recarray'):
                 raise ValueError("cannot specify as_recarray when "
                                  "specifying a multi-index header")
@@ -1184,6 +1186,10 @@ def __init__(self, kwds):
                     raise ValueError("index_col must only contain row numbers "
                                      "when specifying a multi-index header")
 
+        # GH 16338
+        elif self.header is not None and not is_integer(self.header):
+            raise ValueError("header must be integer or list of integers")
+
         self._name_processed = False
 
         self._first_chunk = True
diff --git a/pandas/tests/io/parser/header.py b/pandas/tests/io/parser/header.py
index 1e5fb42b4c1d4..4935fd2cd910a 100644
--- a/pandas/tests/io/parser/header.py
+++ b/pandas/tests/io/parser/header.py
@@ -277,3 +277,12 @@ def test_no_header(self):
         tm.assert_index_equal(df.columns, Index(lrange(5)))
 
         tm.assert_index_equal(df2.columns, Index(names))
+
+    def test_non_int_header(self):
+        # GH 16338
+        msg = 'header must be integer or list of integers'
+        data = """1,2\n3,4"""
+        with tm.assert_raises_regex(ValueError, msg):
+            self.read_csv(StringIO(data), sep=',', header=['a', 'b'])
+        with tm.assert_raises_regex(ValueError, msg):
+            self.read_csv(StringIO(data), sep=',', header='string_header')