Hi Naomi,
>Thanks again for your suggestion. I've updated my FAQ and added your method (3).
>See my test results and I would appreciate more ideas on the topic.

first off as it is a FAQ you should keep the initalizing section
VERY clear to check against the top record. Your codes both are clearly working correctly,
but the reason for this might not be evident for the perhaps tired reader<g>.

*-- init to unequal to save cycles in loop
lcCompMe = iif(testdup.CompMe=="!", "?", "!")

secondly, your benchmark might be quite unfair to 2
and very good for 4, as you have an enormous duplicate rate.

I'ld set up another test run with duplicate frequencies in the area of 1-5%:
here approach 2 can show of his strength, which is the exclusion of all records
being already unique. Also the amount to copy for approach 4 is higher.
Even approach 1 (which prpbably is still the fastest), has to recall more records.

regards

thomas