Ext2 is the old 'tried-and-true' filesystem of Linux. Ext3 adds journaling support on to ext2 (much like ntfs has journaling as opposed to fat32), and I'd asssume any filesystem limits apply to ext3 that also apply to ext2, since it's possible to convert between them. Journaling is a process where any changes you make to files are recorded in a 'journal' first, then the changes are made to the actual file. If the power goes out or something crashes during a write operation, the journal lets the system go back and check the file against the changes to fix any corruption. It does slow down the system (although not very much) as file changes now need to be written several times. Resiserfs is a newer filesystem, also with journaling (I believe its up to version 3 now) that's probably eventually gonna surpass any of the ext types. I believe reiserfs works better with the very large files (being designed at a time when such things are more common). Ext2 is the fastest, because it has no journaling, but the lack of journaling can lead to data loss. Ext3 is considered the most dependable because its been around forever so any bugs in it have probably been rooted out long ago. Reiserfs is the newest, has the most features and so far has been reliable enough for some distros to use as the default. There's more info than you'd ever want to know about file systems, specifically reiser v4 at http://www.namesys.com/v4/v4.html As for file-size, I think ext3 has support for files bigger than 2gb since kernel 2.2.14. I think this might also depend on glibc and the specific programs accessing the files being compiled to access 64 bit integers as well, but that would affect whichever filesystem you use, I think (and I'd assume most programs that need it would have support for that by now). Personally I'm using an 120 gig ext3 partition thats mostly full of 700 mB video files, and I haven't seen any problems, except that every 15th mount or so it checks my disk and boy is it slooooow. Maybe reiserfs fixes this, I'm not sure.