That is a very good idea. What I'm haven't told you is that the slow query was generated from an Excel pivot table via odbc. My user knows that the query is fast on my test box and slow on my prod (using same data sets). I've given him the modified query to resolve the speed issue temporarily, but resolving the difference between the servers has to be my end solution. I'm hoping this RedGate you suggested can help me find out what is different.

I was thinking of running a dif on the two text execution plans but I don't know if I would understand that. That will be my next approach after RedGate. The execution plans on test and production should be identical, if not then I'm on to something. If they are identical then I will go crazy.