Select Many-to-Many with more than one count?

[eluser]Lockzi[/eluser]
Hi,

I would love to do the following:

select all rows where 'another_id' has more than one key_id but that key_id has no other.


In this example:

Code:

relation_id key_id another_id
1           1      5
2           3      5
3           7      6
4           5      9
5           21     13
6           5      19
7           5      26
8           3      26

PSUDO CODE

Code:

SELECT * FROM this_table WHERE another_id = 5 AND (each key_id in result DOESNOT have other another_id)

The results should be:

Code:

1           1      5

But NOT

Code:

2           3      5

Since the key_id 3 also has another_id = 26:

Code:

8           3      26

I've personally been facing this problem for a couple of days now.
What I've found so far is subqueries (mainly from here http://shawnmccool.com/2009/09/18/using-...ubqueries/, but I just can't make it work the way I want it.

Am I going about this all wrong?

Greatly appreciated input,
Lockzi