I've always used two steps ... search on indexed field then constrain on unindexed! Great question!
LaRetta points out a way to do this. When FileMaker does a find based on multiple fields, it does it as one step.. So if a field in the find is unstored, it will slow the find down. If you are doing the find manually or scripted, you can do it in two steps. Do the find on the stored field first and narrow the record count. Then constrain the found set with a second find/ constrain found set.