>>Frank,
>>
>>Datetimes are accurate to the millisecond but we can only access the seconds. This si why you are getting the rounding error. Better than round() in this situation is ceiling() as in;
>>
>>lnSecs = CEILING(DT1 - DT2)
>
>Why it's better?

IMO, CEILING() is the math operation to return the next higher integer, round() is the math operation to round off either up or down to a fixed number of decimals. Ceiling is the more appropriate math opreation for this situation as the desired result is always the next higher integer never a non-integer or a lower value.