>How big is the table in bytes? I did a similar test on a table with 4 fields (recsize of about 50 bytes) with 1,000,000 records. With deleted on and deleted off. There was a difference in some commands with sql that was dramatic percentwise (0.03 sec in deleted off and .5 sec with deleted on) but 0.5 sec is still ok by my standards. This is with the first 1000 records deleted and a simple Rushmore optimizable sql.

The problem is not with SQL.

LOCATE and SCAN are not optimizable anymore.

Yes, SET OPTIMIZE is the default, which is ON.