Here's a solution structured as a subroutine; the "main" portion just uses your example data, and when run at the command line, it uses the first argument after the script name as the value of "$x":

#!/usr/bin/perl

$num{'15'}=1;
$num{'20'}=2;
$num{'25'}=3;
$num{'30'}=4;

$x=shift;  # take value from @ARGV

print value_sought( $x ), $/;

sub value_sought
{
    my $x = shift;  # take value from @_
    foreach my $k ( sort { $a <=> $b } keys %num )
    {
        return $num{$k} if ( $x < $k );
    }
    return "$x is larger than available num keys";
}
[download]

But, as Abigail suggested in an earlier reply, if you're doing frequent searches over a large set of hash keys, you may want to optimize this a bit -- sort the keys into an array, check the value of $x against the middle element of that array, then the middle of the upper half or lower half as appropriate, until you hone in on the right key value; in the majority of cases, the correct element will be found in fewer iterations this way.

(If the set of hash keys is fairly static -- you do a lot of searches over the same set of hash keys -- then having the array will also help by not having to sort the hash keys every time you do a search.)