Gail, thanks for prompt response.

I think you are right about IO stack and not hard drive. Customer created a test DB from backup of prod DB made after problem arose. We get same error in test DB.

The subject DB is created and maintained by a ERP system that does not use database referential integrity. Another option I thought of is:

a.Script all indexes from offending table PJRate

b.Create backup of PJRate (select * into xPJRate from PJRate)

c.Create indexes on xPJRate

d.Drop table PJRate

e.Rename xPJRate to PJRate.

Since the DBCC CHECKDB messages show the error on a clusteded index, my alternative approach may not be any betty than dropping indexes on offending table and then rebuilding.

Thanks.