>>Cetin,
>>
>>Just an update. I've retested with Sergey's and Rick's solution and I am now seeing similar results. Your patch for the random day value cured most of my ills. "Garbage in, garbage out." Thanks again.
>
>Be carefull though. I could reproduce differences up to 28 days. All dates are related with 2/29 and 1/3 (who invented this calendar:)
>Cetin

Agreed. I'm seeing 28-30 in the cases when the DOD falls in the 28-31 day range of the month. I suspect this is the GOMONTH() problem you and Rick referred to. I'm just rescanning the file and tweaking the DOB with the variance. The client is estatic. I'm just happy about it ;-)