I would strongly suggest comparing the contents of the hold files HOLDACCT, HOLDJOBS, HOLDMNTH and HOLDYEAR, from the run when it is first and when it is second.

quote:

1ST: HOLDACCT DISK 66 V D:\ibi\srv76\dm\edatemp\ts000079\holdacct.ftm
2ND: HOLDACCT DISK 25 V D:\ibi\srv76\dm\edatemp\ts000079\holdacct.ftm

I didn't check them all but the record lengths of these two are different between your two different posts of filedef results. You need to investigate why that is.

You might want to consider doing an APP HOLD to two different permanent locations, moving the 'bad' fex around so that you can look at and compare the contents of the files you are using to test against in your WHERE clauses. I am almost certain that is where the problem is.

I would try moving all of your hold file logic used to create the hold files for your WHERE clauses outside of the PDF OPEN / PDF CLOSE code stream. The fact that the report runs fine in second/last position when the WHERE clauses are removed makes me suspicious not of the HOLD files, but the MODIFY logic that creates them.