It looks like the initial LIKE is doing the bulk of the filtering – could this be where the performance gains are coming from?

From what I know, MySQL will optimize the query before running it and will get the first result set using the LIKE and then use the 2nd and 3rd REGEXP’s to filter the result set further before returning it (but I could be wrong)

Paul

Looks good, although I think in your specific example the

AND field NOT LIKE ‘%material%’

portion will exclude some results that you wouldn’t necessarily want to exclude. e.g. if field was “build materialx” it would be excluded in your optimized but not in the original regex version.