The precedence problem has been covered. As for idiomaticity (?), how about:

c:\@Work\Perl>perl -wMstrict -le
"for ('filesystem          kbytes   used     avail    capacity  mounte
+d on',
      '/dev/dsk/c0b0t0d0s4 12209400 5496486  6712914  46%       /home'
+,
   ) {
   printf qq{'%s' -> '%s' \n}, $_, parse_capacity($_);
   }
 ;;
 sub parse_capacity {
   my ($df_output) = @_;
   return $df_output =~ m[(\d+)%\s+/home$] ? $1 : 'Match Error';
   }
"
'filesystem          kbytes   used     avail    capacity  mounted on' 
+-> 'Match Error'
'/dev/dsk/c0b0t0d0s4 12209400 5496486  6712914  46%       /home' -> '4
+6'
[download]

Update: Actually, if the structure of a data record really is that simple, it's not that much more effort to go ahead and parse the whole thing:

c:\@Work\Perl>perl -wMstrict -le
"use Data::Dump qw(dd);
 ;;
 for ('filesystem          kbytes   used     avail    capacity  mounte
+d on',
      '/dev/dsk/c0b0t0d0s4 12209400 5496486  6712914  46%       /home'
+,
   ) {
   dd $_, parse_all($_);
   }
 ;;
 sub parse_all {
   my ($df_output) = @_;
   ;;
   my $match =
   my ($fsys, $kb, $used, $avail, $cap, $mtd) = 
   $df_output =~ m{
     \A (.+) \s+ (\d+) \s+ (\d+) \s+ (\d+) \s+ (\d+)% \s+ (.+) \z
     }xms;
   ;;
   return { error => 'Match Error' } unless $match;
   ;;
   return {
     fsys       => $fsys,
     kb         => $kb,
     used       => $used,
     avail      => $avail,
     cap        => $cap,
     mounted_on => $mtd,
     };
   }
"
(
  "filesystem          kbytes   used     avail    capacity  mounted on
+",
  { error => "Match Error" },
)

(
  "/dev/dsk/c0b0t0d0s4 12209400 5496486  6712914  46%       /home",
  {
    avail => 6712914,
    cap => 46,
    fsys => "/dev/dsk/c0b0t0d0s4",
    kb => 12209400,
    mounted_on => "/home",
    used => 5496486,
  },
)
[download]