>>Recreate the indeces.
>
>Hey just for the record, I didn't have to go there. I really think this is some sort of post SP bug perhaps.
>
>I would of had to wait awhile to get an exclusive lock on my source files to recreate the indexes. So I went ahead and rewrote the legacy code's SCAN/ENDSCAN (to build static copies of my source tables) logic into straghtforward queries w/subqueries then appended my results into the replicated tables. Worked fine! Worked real fast too of course. No Invalid Seek Offset error, and no C5.

Glad to see you got it fixed. You going to do more research into the problem?