>Actually, the amount is capped at 2gb in GetDiskFreeSpace(), and that's what throws it off. It may be coincidental that it's right. As I mentioned earlier in this thread, GetDiskFreeSpaceEx() wasn't available until Win95 OSR2.

If disk space was what was needed, I would say use the new function. However, GetDiskFreeSpaceEx does not provide the same functionality as its lesser cousin.

The point that I was trying to make was that while the reported free disk space may be skewed on Win9X machines (or on all machines), the cluster size was reported correctly on an NT machine using both GetDiskFreeSpace diirectly and Sys(2022) which uses that function internally.

Once again, support for the WinAPI on an NT machine is proved to an order of magnitude better (that's my pet phrase this week *s*) than on its "not really a 32bit OS" cousin Win9X.