I would also add List::Moreutils, which has a minmax function using an improved algorithm to evaluate min and max at once:

#!/usr/bin/perl
use strict;
use warnings;
use List::Util qw( min max reduce );
use List::MoreUtils qw( minmax );
use Benchmark qw( cmpthese );

my @array = map { rand 100 } 1 .. 1_000_000;

my $count = 0;
cmpthese -10, {
   util_min_max => sub {
      my $min = min(@array);
      my $max = max(@array);
   },
   util_reduce => sub {
      my ( $min, $max ) = @{ (reduce {
         my $r= ref $a ? $a : [($a) x 2];
         if ($b < $r->[0]) {
            $r->[0]= $b;
         } elsif ($b > $r->[1]) {
            $r->[1]= $b;
         }
         $r
      } @array) };
   },
   moreutils_minmax => sub {
      my ( $min, $max ) = minmax(@array);
   },
};

__END__

                    Rate      util_reduce     util_min_max moreutils_m
+inmax
util_reduce      0.846/s               --             -92%            
+ -96%
util_min_max      10.5/s            1146%               --            
+ -45%
moreutils_minmax  19.2/s            2169%              82%            
+   --
[download]

I added the reduce method because I did not know if it would perform better, but the result does not surprise me given the bunch of function calls that are involved. Anyway, this is what benchmarks are for :)