If I understand you well, your input line looks like '3256 nothing special 356216'. In this case it's enough to match digit, non-digit and digit again...

#!/usr/bin/perl
while (<>) {
        print if m/(\d+)\D+(\d+)/;
        print "$1\n$2\n";
}
[download]

The output is:

tm@norad:~$ ./regex_test
123 testing... testing 456
123 testing... testing 456
123
456
[download]

UPDATE: after reading next replies...

1. .+ isn't the best notation - it shouldn't catch next digits after (\d+) but it's not specially safe AIMO, especially makes the code more difficoult to understand
2. this 'to', whitespaces or other words you mentioned in your problem are covered with \D+ which IMvHO is more universal becouse of catching anything except digits

Greetz, Tom.