>>It's a known problem. Datetime stores the milliseconds and if you try to use in direct calculation that causes headache. If hour portions modulus 3 were equal (mins 0, secs 0) you'd get exact numbers (ie: 9 PM and 6 AM or 7 PM and 10 AM). int(ltEnd-ltStart)
>>Cetin
>
>Is there a way to retrieve those milliseconds?
>
>Hilmar.

There is but not via usual VFP functions :( If you could access the table lowlevel you could get the data sotored in 8 bytes. First 4 are integer portion storing the dates passed since (1899/12/30 - not sure it might be 1900 based but possibly this one), the other four bytes fractions of a day. However it's been long I didn't hack it :) Double check.

Update: Maybe you could get it subtracting the datetime value from ctot(ttoc(dtValue)) divided by 1.0000
Cetin