Good point. Here is a better version that does not have to define this list first and does not make any assumptions about the size values.

#!/usr/bin/perl
use strict; 
use warnings; 
use Data::Dumper;

my $keep = 5;
my @keeper;

while (<DATA>) {
    my @vals  = split;
    next unless @vals; # possibly more sophisticated test
    my $score = pop @vals;
    addlist ([$score, \@vals], \@keeper);
}

print Dumper \@keeper;

# slide up the list knocking down existing records
# until we no longer have the biggest
sub addlist {
    my ($rec, $list) = @_;
    for my $i (0 .. $keep-1) {
        if (not $list->[$i] or $rec->[0] > $list->[$i]->[0]) {
            $list->[$i-1] = $list->[$i] if $i;
            $list->[$i]   = $rec;
        }
        else {
            last;
        }
    }
}

__DATA__
0 5 3 7   0.97
1 3 4 8   0.18
1 4 4 8   0.21
1 7 4 4   0.22
1 8 4 8   0.52
1 9 4 1   0.16
1 5 1 8   0.18
1 5 2 9   0.72
1 5 5 8   0.32
1 5 6 6   0.17
1 5 7 4   0.52
1 5 9 3   0.92
1 5 4 8   0.99
6 8 4 6   0.54
[download]

Cheers,
R.