>>>I think INSERT INTO still has to lock the header momentarily while it updates the record counter. More than one INSERT INTO by users in multitask can create the error.
>>
>>But, I thought this line was taking care of that?
>>
>>SET REPROCESS TO 2 SECONDS
>
>2 seconds does seem like enough time for header update.
>
>If the server is slow, and the 2 seconds elapses, then the error is still generated. Maybe that is why it only occurs infrequently.

On re-reading the SET REPROCESS TO ... help entry, I noticed that

"If an ON ERROR routine is in effect ... the ON ERROR routine takes precedence over additional attempts to lock the record or file..."

Could this have contributed to the error? It sounds as if even one lockout would kick you into the error routine.

Barbara