Personally, I'd go with tbone1's idea: "ls -t" will list directory contents in reverse-cronological order (newest file first), while "ls -t -r" will do the opposite (oldest file first) -- assign the back-tick output to an array and you're done.

If you want something that will work on systems that don't support "ls -t", then I think you should start with a hash, keyed by file name, with file ages as the values -- it involves doing a stat on each file just once (rather than multiple times within a sort block, as suggested in ccn's reply). Something like this:

my %files;

opendir( D, "." );   # or some specific path?
-f and $files{$_} = -M _ for ( readdir D );
closedir D;

my @newest_first = sort { $files{$a} <=> $files{$b} } keys %files;
[download]

But if you're just running you code "in shop" (as opposed to including it in a package for open distribution), using "ls" with back-ticks is fine.