diff --git a/python/umysql.c b/python/umysql.c
index 4cc7d47..1fde43b 100644
--- a/python/umysql.c
+++ b/python/umysql.c
@@ -621,8 +621,11 @@ int API_resultRowValue(void *result, int column, UMTypeInfo *ti, char *value, si
     case MFTYPE_MEDIUM_BLOB:
     case MFTYPE_LONG_BLOB:
     case MFTYPE_BLOB:
-      // Fall through for string encoding
-      valobj = PyString_FromStringAndSize( (const char *) value, cbValue);
+      if (ti->flags & MFFLAG_BINARY_FLAG) {
+        valobj = PyString_FromStringAndSize( (const char *) value, cbValue);
+      } else {
+        valobj = DecodeString (ti, value, cbValue);
+      }
       break;
 
       //PyString family
diff --git a/tests/tests.py b/tests/tests.py
index d1345de..544cdb6 100644
--- a/tests/tests.py
+++ b/tests/tests.py
@@ -659,7 +659,37 @@ def testCharsets(self):
             self.assertEquals(result, expected)
 
         cnn.close()
-    
+
+    def testTextCharsets(self):
+        aumlaut_unicode = u"\u00e4"
+        aumlaut_utf8 = "\xc3\xa4"
+        aumlaut_latin1 = "\xe4"
+
+        cnn = umysql.Connection()
+        cnn.connect (DB_HOST, 3306, DB_USER, DB_PASSWD, DB_DB)
+
+        cnn.query("drop table if exists tblutf")
+        cnn.query("create table tblutf (test_mode TEXT DEFAULT NULL, test_utf TEXT DEFAULT NULL, test_latin1 TEXT) ENGINE=MyISAM DEFAULT CHARSET=utf8")
+
+        # We insert the same character using two different encodings
+        cnn.query("set names utf8")
+        cnn.query("insert into tblutf (test_mode, test_utf, test_latin1) values ('utf8', _utf8'" + aumlaut_utf8 + "', _latin1'" + aumlaut_latin1 + "')")
+        
+        cnn.query("set names latin1")
+        cnn.query("insert into tblutf (test_mode, test_utf, test_latin1) values ('latin1', _utf8'" + aumlaut_utf8 + "', _latin1'" + aumlaut_latin1 + "')")
+
+        # We expect the driver to always give us unicode strings back
+        expected = [(u"utf8", aumlaut_unicode, aumlaut_unicode), (u"latin1", aumlaut_unicode, aumlaut_unicode)]
+
+        # Fetch and test with different charsets
+        for charset in ("latin1", "utf8", "cp1250"):
+            cnn.query("set names " + charset)
+            rs = cnn.query("select test_mode, test_utf, test_latin1 from tblutf")
+            result = rs.rows
+            self.assertEquals(result, expected)
+
+        cnn.close()
+     
 if __name__ == '__main__':
     from guppy import hpy
     hp = hpy()