I am tending to agree with you on available RAM. When suspending the program during heavy execution (and therefore memo bloat), there is a significant pause with a lot of disk activity while the sytem clears it's cache and finishes writing the temp files (you can see them grow significantly at that time). I keep coming back to the one fact, though, that despite RAM, cache, or anything else, I hate to think that exclusively used tables experience memo bloat. It kind of rocks the foundation of a core belief that I've had since FoxPro was invented!