I inherited a piece of code that had a large number of replace statements after using several large select statements to build cursors and did some calculations along the way. It took 15 minutes to produce a report on a dataset with over 300,000 records. I replaced all of the replace statements I could with scatter and gathers and combined the other replace statements ( one that entailed some calculations ) into one replace statement ( one statement excutes faster than many ). This sped up my process 4.5 minutes. I also replaced all the locate for ... with seek after indexing the cursors I produced. The time dropped to just under 3 minutes. Just an idea.