diff --git a/termstr.py b/termstr.py
index 9e816d2..5ac70f4 100644
--- a/termstr.py
+++ b/termstr.py
@@ -171,9 +171,16 @@ class TermStr(collections.UserString):
         return self._split_style(self.data.split(sep, maxsplit), len(sep))
 
     def splitlines(self, keepends=0):
-        # FIX. Fails when a line seperator isn't one character in length.. \r\n
-        sep_length = 0 if keepends else len("\n")
-        return self._split_style(self.data.splitlines(keepends), sep_length)
+        lines_with_ends = self.data.splitlines(keepends=True)
+        lines_without_ends = self.data.splitlines()
+        result_parts = lines_with_ends if keepends else lines_without_ends
+        result = []
+        cursor = 0
+        for line, line_with_end in zip(result_parts, lines_with_ends):
+            style_part = self.style[cursor:cursor+len(line)]
+            result.append(self.__class__(line, style_part))
+            cursor += len(line_with_end)
+        return result
 
     def capitalize(self):
         return self.__class__(self.data.capitalize(), self.style)
diff --git a/termstr_test.py b/termstr_test.py
index c24e123..8d2eaa4 100755
--- a/termstr_test.py
+++ b/termstr_test.py
@@ -96,8 +96,11 @@ class TermStrTests(unittest.TestCase):
         self.assertEqual(foo.join(["C"]), TermStr("C"))
         bar = TermStr("bar", bold_style)
         self.assertEqual((foo + "\n" + bar).splitlines(), [foo, bar])
+        self.assertEqual((foo + "\r\n" + bar).splitlines(), [foo, bar])
         self.assertEqual((foo + "\n" + bar).splitlines(keepends=True),
                          [TermStr("foo\n"), bar])
+        self.assertEqual((foo + "\r\n" + bar).splitlines(keepends=True),
+                         [TermStr("foo\r\n"), bar])
         self.assertEqual(foo.ljust(5), foo + TermStr("  "))
         self.assertEqual(foo.rjust(5), TermStr("  ") + foo)
         self.assertEqual(TermStr("FOO").lower(), foo)