>In my original tests I used to get this error on every second or third attempt. I'm not getting this error at all now no matter how many times I load different files. So, with a big % of certanity I can say that the error has gone. I can not detect what exactly eliminated the error, because I made a lot of changes. The only thing that remains common in the two programs is the business logic, the program flow and table aliases...

Sounds like not enough SourceSafe use <g>. But if you were asked to isolate the cause (as it might be relevant to al the other errors you get) you could easily zero into it by analogous technique to binary search - make always half of the difference active. The big win is that you already have one working and one failing solution.

regards

thomas