diff --git a/src/pyodbcmodule.cpp b/src/pyodbcmodule.cpp
index f6939e14..c085fcda 100644
--- a/src/pyodbcmodule.cpp
+++ b/src/pyodbcmodule.cpp
@@ -453,29 +453,37 @@ static PyObject* mod_drivers(PyObject* self)
     if (henv == SQL_NULL_HANDLE && !AllocateEnv())
         return 0;
 
-    PyObject* result = PyList_New(0);
+    Object result(PyList_New(0));
     if (!result)
         return 0;
 
-    SQLCHAR szDriverDesc[SQL_MAX_DSN_LENGTH];
+    SQLCHAR szDriverDesc[500];
     SWORD cbDriverDesc;
-    SQLCHAR szDesc[200];
-    SWORD cbDesc;
-
-    SQLUSMALLINT nDirection = SQL_FETCH_FIRST;
+    SWORD cbAttrs;
 
     SQLRETURN ret;
+    SQLUSMALLINT nDirection = SQL_FETCH_FIRST;
 
     for (;;)
     {
         Py_BEGIN_ALLOW_THREADS
-        ret = SQLDrivers(henv, nDirection, szDriverDesc,  _countof(szDriverDesc),  &cbDriverDesc, szDesc, _countof(szDesc), &cbDesc);
+        ret = SQLDrivers(henv, nDirection, szDriverDesc, _countof(szDriverDesc), &cbDriverDesc, 0, 0, &cbAttrs);
         Py_END_ALLOW_THREADS
+
         if (!SQL_SUCCEEDED(ret))
             break;
 
-        if (PyList_Append(result, PyString_FromString((const char*)szDriverDesc)) != 0)
+        // REVIEW: This is another reason why we really need a factory that we can use.  At this
+        // point we don't have a global text encoding that we can assume for this.  Somehow it
+        // seems to be working to use UTF-8, even on Windows.
+        Object name(PyString_FromString((const char*)szDriverDesc));
+        if (!name)
+            return 0;
+
+        if (PyList_Append(result, name.Get()) != 0)
             return 0;
+        name.Detach();
+
         nDirection = SQL_FETCH_NEXT;
     }
 
@@ -485,7 +493,7 @@ static PyObject* mod_drivers(PyObject* self)
         return RaiseErrorFromHandle("SQLDrivers", SQL_NULL_HANDLE, SQL_NULL_HANDLE);
     }
 
-    return result;
+    return result.Detach();
 }