diff --git a/Doc/library/stringio.rst b/Doc/library/stringio.rst
--- a/Doc/library/stringio.rst
+++ b/Doc/library/stringio.rst
@@ -82,10 +82,7 @@ instead.
    those cases.
 
    Unlike the :mod:`StringIO` module, this module is not able to accept Unicode
-   strings that cannot be encoded as plain ASCII strings.  Calling
-   :func:`StringIO` with a Unicode string parameter populates the object with
-   the buffer representation of the Unicode string instead of encoding the
-   string.
+   strings that cannot be encoded as plain ASCII strings.
 
    Another difference from the :mod:`StringIO` module is that calling
    :func:`StringIO` with a string parameter creates a read-only object. Unlike an
diff --git a/Lib/test/test_StringIO.py b/Lib/test/test_StringIO.py
--- a/Lib/test/test_StringIO.py
+++ b/Lib/test/test_StringIO.py
@@ -135,6 +135,28 @@ class TestcStringIO(TestGenericStringIO)
         self.assertEqual(f.getvalue(), '\x00\x01\x02')
 
 
+    def test_unicode(self):
+
+        if not test_support.have_unicode: return
+
+        # The cStringIO module converts Unicode strings to character
+        # strings when writing them to cStringIO objects.
+        # Check that this works.
+
+        f = self.MODULE.StringIO()
+        f.write(unicode(self._line[:5]))
+        s = f.getvalue()
+        self.assertEqual(s, 'abcde')
+        self.assertEqual(type(s), types.StringType)
+
+        f = self.MODULE.StringIO(unicode(self._line[:5]))
+        s = f.getvalue()
+        self.assertEqual(s, 'abcde')
+        self.assertEqual(type(s), types.StringType)
+
+        self.assertRaises(UnicodeEncodeError, self.MODULE.StringIO,
+                          unicode('\xf4', 'latin-1'))
+
 import sys
 if sys.platform.startswith('java'):
     # Jython doesn't have a buffer object, so we just do a useless
diff --git a/Misc/NEWS b/Misc/NEWS
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -66,6 +66,10 @@ Core and Builtins
 Library
 -------
 
+- Issue #1548891: The cStringIO.StringIO() constructor now encodes unicode
+  arguments with the system default encoding just like the write() method
+  does, instead of converting it to a raw buffer.
+
 - Issue #9168: now smtpd is able to bind privileged port.
 
 - Issue #12529: fix cgi.parse_header issue on strings with double-quotes and
diff --git a/Modules/cStringIO.c b/Modules/cStringIO.c
--- a/Modules/cStringIO.c
+++ b/Modules/cStringIO.c
@@ -661,7 +661,11 @@ newIobject(PyObject *s) {
   char *buf;
   Py_ssize_t size;
 
-  if (PyObject_AsReadBuffer(s, (const void **)&buf, &size)) {
+  if (PyUnicode_Check(s)) {
+    if (PyObject_AsCharBuffer(s, (const char **)&buf, &size) != 0)
+      return NULL;
+  }
+  else if (PyObject_AsReadBuffer(s, (const void **)&buf, &size)) {
     PyErr_Format(PyExc_TypeError, "expected read buffer, %.200s found",
                  s->ob_type->tp_name);
     return NULL;