After reviewing the CPAN Perldoc, it looks as though it most certainly should be returning a list back from the method, therefore you should be using it in the correct context. However, it does also note:

The method for specifying link criteria is the same as in find_link().

And, when reviewing the find_link() method, it looks as though it's implemented such as:

  my $agent = WWW::Mechanize->new();
  my $link = $agent->find_link( text => "some text", url_regex => qr/s
+omelink\.com/ );
[download]

Using this as an example, I would try re-factoring your code to read:

#!/usr/bin/perl 
use strict;
use www::mechanize;

my $agent = WWW::Mechanize->new();

my @urls;

my $regex = 'results';

my $url ='http://www.google.co.uk/search?hl=en&safe=off&q=intitle%3A%2
+2football+scores%22+inurl%3Aresults&meta=cr%3DcountryUK%7CcountryGB';

  $agent->get($url); 
  
  my @urls=$agent->find_all_links(text => "some text", url_regex => qr
+/somelink\.com/);

print stdout "@urls\n";
[download]

For the parameters of the method, you may want to take a look at the Perldoc for the class you are trying to use. There are a number of parameters you can supply, though they must be supplied as defined. Therefore, in your example, it would probably only search for any links that look like '

http://www.google.co.uk/search?hl=en&safe=off&q=intitle%3A%2
+2football+scores%22+inurl%3Aresults&meta=cr%3DcountryUK%7CcountryGB
[download]

', which would probably be none in the results.

One last (I swear) thing you might want to take a look at is the Google API. Although it is a SOAP based API, it is somewhat easy to use and there are already Perl modules in CPAN that act as wrappers for the API.

Also, maybe take a look into WWW::Scraper::Google and DBD::Google. Personally, I have not used them nor looked at the source to see what they are doing. However, if stable, they are most likely doing the same thing you are trying to accomplish.

Good Luck!