>select * from table1;
>where keyfield NOT IN;
>(select keyfield from table2)

>Any ideas for something faster?

I don't think I saw this option. Since the sub-query isn't optimizable, this might be faster if the records of table2 in table1 is a small percentage or the whole.

First pull the records from table2 that exist in table1. This should be fully optimized.

SELECT table2.keyfield;
    FROM Table2 JOIN Table1 
        ON table2.keyfield-table1.keyfield;
    INTO CURSOR temp

SELECT *
    FROM table1
    WHERE keyfield NOT IN;
        (SELECT keyfield from temp)

hth