Unable to duplicate slow query log rows_examined

I have a query that according to my slow query log it's a bit slow....

Query_time: 8.408943  Lock_time: 0.000119 Rows_sent: 1  Rows_examined: 2911766

BUT, when I run the query with EXPLAIN in front of it, I don't get the same results...

id select_type table type possible_keys key key_len ref rows Extra
1 PRIMARY forum range PRIMARY PRIMARY 4 NULL 3 Using where; Using temporary; Using filesort
1 PRIMARY catego开发者_StackOverflow社区ry ref PRIMARY,forum_id forum_id 4 source_forum.forum.id 2  
1 PRIMARY board ref PRIMARY,category_id category_id 4 source_forum.category.id 4 Using where
1 PRIMARY topic ref PRIMARY,board_id board_id 4 source_forum.board.id 58  
1 PRIMARY post ref PRIMARY,topic_id,trash topic_id 4 source_forum.topic.id 16 Using where
3 DEPENDENT SUBQUERY post index topic_id created 4 NULL 1 Using where
2 DEPENDENT SUBQUERY group_assoc ref board_id,group_id board_id 4 source_forum.board.id 4 Using where

The highest row count there being used is 56...

update

My query:

SELECT
   COUNT(id) AS num
FROM (
    SELECT topic.*,
      (SELECT created FROM post WHERE topic_id = topic.id ORDER BY created DESC LIMIT 1) AS lastpost
    FROM topic
    WHERE board_id = 6 AND
    NOT EXISTS( SELECT id FROM topic_read_assoc WHERE topic_id = topic.id AND member_id = 489 )            
    ) tab
WHERE last_post_time > 1288032259;

explain extended

id select_type table type possible_keys key key_len ref rows filtered Extra
1 PRIMARY <derived2> ALL NULL NULL NULL NULL 440 100.00 Using where
2 DERIVED topic ref board_id board_id 4   429 100.00 Using where
4 DEPENDENT SUBQUERY topic_read_assoc ref topic_id,member_id topic_id 4 source_forum.topic.id 6 100.00 Using where
3 DEPENDENT SUBQUERY post index topic_id created 4 NULL 1 1600.00 Using where

What does filtered mean?

---

Can you post the query and the SHOW CREATE TABLE statements too?

58 indeed is not high, but you are using temporary tables AND a filesort. And since all your types are ref and not eq_ref, you have to multiply those values: 3*2*4*58*16*1*4 = 89k rows joined (see the rows examined - some tables probably got scanned entirely - see the USING WHERE comment)

Can you post an EXPLAIN EXTENDED and SHOW WARNINGS?