I was hoping to find a nice long thread about this topic so I could hear some of the pros/cons and experiences others have had with this method.

We are currently designing a new system where the normal problems of using an identity field are not an issue (no worries about migration from other systems, all inserts of normalized tables through sProcs so we can do them in a single trip, etc).

That being the case, do we really gain or lose anything by using a UniqueIdentifier field vs Identity (other than the increase in field size)? For initial testing, I do find it easier using the identity field as I can immediately see issues / relationships vs a list of guids.

Any comments or links to other articles / discussions on this would be appreciated.