Hey Sergey,

You are right, this is a much better way to do this. I have found my "anomoly", they sent us the exact same file as last month. Not sure why that would have the effect it did on that routine, but using your's and Mike's example showed me that right away. Thanks for your help.
Kevin


>Maybe select would be faster in this case?

SET TALK ON NOWINDOW
>SELECT * FROM new ;
>  WHERE new.ssn + new.holder_num + DTOS(new.date_rem) NOT IN ( ;
>    SELECT old.ssn + old.holder_num + DTOS(old.date_rem) FROM old ) ;
>  INTO CURSOR crsAdditions
>SET TALK OFF

>
>>It's about the same. This is a data cd I get from the state every month. The processing we are doing is to figure out what they have added in the past month. So basically every time I run this, it takes the file from 400,000 records to usually a couple or few thousand. So to answer your question, almost the entire file is deleted every month. KP
>>
>>>>No, the routine before this one alters the table and packs it. Also, I just ran the routine using last months data (approx the same size) and this routine ran in about 10 minutes. Very frustrating!!!
>>>>
>>>Hi Kevin,
>>>
>>>Maybe this month many more records have to be deleted? How many records were deleted last month and how many will be deleted this month?