The problems with this method are documented in 'perldoc DBI'. It is probably more reliable than I would be, but to sum it up, $sth->rows is mostly for do() commands like INSERT, DELETE, etc, so that you know how many rows were affected (as those queries don't really have any other results). I also suspect that DBI functionality, although it is supposed to provide a uniform interface, can vary somewhat based on which type of server is used. There might be a MySQL-specific answer to this question, but it probably wouldn't be very portable-- besides, doing COUNT(*) or simply fetching all the data (since, in this case, only a single row seems to be expected), is just going to be quicker than any other solution.