You're following ->prepare, ->execute because that's how the DBI API is designed. A DBD is free to do as much or as little work in any of the two steps. I think DBD::mysql merely stores the SQL string in ->prepare and does all the checking etc. in ->execute. You can switch to just using ->selectall_arrayref, but that way you lose easy access to caching. Also, should you ever switch to a different database, it might do more work during ->prepare and thus that phase might raise errors.