runrig, I feel like an idiot :)

Below I offered what I thought was a different perspective on this problem, then I realized I had redone your solution. Ooops.

I would, however, suggest that your solution could be improved by using split and join as follows:

#!/usr/local/bin/perl -l -w

use strict;

my $str = "SAdBoy";

$str =~ s/($str)/fix_case($1, 'badgirl')/eig;

print $str;

sub fix_case {
 my ($match_word, $replace_word) = @_;
 my @rep = split //, $replace_word;
 my $i = 0;
 for (split '', $match_word) {
  $rep[$i] = $_ eq lc($_) ? lc($rep[$i++]) : uc($rep[$i++]);
 }
 return join '', @rep;
}
[download]

I also tossed in a lc() so that capitalization in the $replace_word doesn't "contaminate" the pattern in the $match_word.

I offer this because I am of the impression that using substr that often is somewhat expensive.