Jefterwil, to do this neatly and cleanly in SQL you need subselects, which are not supported in MySQL < 4.0, and are buggy in the versions that support them when you're referencing a single table.
You might be able to fake it by using a temporary table and a join, but I haven't tried that route.
I ran into a similar problem recently using mysql 3.x, and I resorted to solving it with brute force and PHP. I implemented the equivalent of DISTINCT and LIMIT in PHP.
You have to build an array of stuff you've already seen and don't want to see again (while looping through your first query).
Then remove LIMIT from your second query -- or set it reasonably high.
Loop through the second result set by using an explicit counter variable.
Check each retrieved ID to see if it's in the saved array (there's your DISTINCT) . If it's not, do your output stuff and increment the counter (there's your LIMIT).
I think that's what stevesweetland is trying to do with his example.