henry1951 wrote: I chose them for the absolute certainty that there would be no collisions in IDs Actually it's not an absolute certainty. It's just that the chance of a collision is extremely small. Be careful using GUIDs as PKs. Significant performance issues can arise (both time and space are wasted) when the clustering index is a GUID and the default in SQLServer is to create a clustering index on the PK. You can get around this by using the newsequentialid()function or by generating your own sequential GUIDs. See this article[^]for a discussion of the issue.