>>>I have found that under Windows NT you have to execute
>>>SELECT tablename
>>>=RLOCK()
>>>UNLOCK
>>>after a table update in order to view the "current" information.
>>
>>Larry;
>>
>>I tried your idea and got no change in the results. The machine I am on is a win98 os. Requerying twice is still the only thing that has found the records.
>
>How about issuing a SKIP after you delete, but before the requery?

Larry; The SKIP worked. The idea of the record pointer being over a deleted record causing the strange behavior must of been correct. Locate didn't do it, I am not sure why, but probably a cdx thing. I am glad that someone else was able to recreate this, a programmer's sanity can be a delicate entity.:)

thanx for you input