Hope I'm not telling you what you already know, but if you need to pull the column names out of the table, you could go with opening up a filehandle and splitting. But I agree with the above: using DBD::CSV will scale nicely if you expect your app and data to grow beyond the usefulness of a flat text db.