>My problem
>
>I have two equal in every respect query. But querys speed are very different.
>I have looked at fox.wiki, ut thread and other sources to solve this problem.
>
>I know that, rushmore optimization technique cannot optimize the query for some situation.
>But my query has very basic structure that is easily parsed by rushmore
>
>
>Main Table
>-----------
>Abone : 715.317 records (Abone table has index on ekler and hurun fields)
>
>
>Child Tables
>-----------------------
>Hurun : 20 records (Hurun table has index on id field)
>Ekler : 16 records (Ekler table has index on id field)
>
>
>All the fields (abone.hurun, abone.ekler, ekler.id and hurun.id) are int type and indexed
>Index collate ise turkish and alt tables has türkish code page. && cpdbf("abone") returns 1254
>
>
>--------------------------
> First query
>--------------------------
>

>select * ;
>  from abone, ;
>       hurun, ;
>       ekler ;
> where abone.hurun  = hurun.id ;
>   and abone.ekler  = ekler.id ;
>   and hurun.exp like "Aksi%"
>

>
>Result : selected 309 records in 6.48 seconds
>
>
>---------------------------
> Second query
>---------------------------
>
>

>
>select * ;
>  from abone, ;
>       hurun, ;
>       ekler ;
> where abone.hurun  = hurun.id ;
>   and abone.ekler  = ekler.id ;
>   and ekler.exp like "Arkd%"
>

>
>
>Result : Selected 1155 records in 0.03 seconds

Ali,
Woudl you try like this:

set ansi off && default
select * ;
  from abone  ;
  inner join hurun on abone.hurun  = hurun.id ;
  inner join ekler on abone.ekler  = ekler.id ;
 where ekler.exp = "Arkd"

Is there a reason you're using Turkish collate. I suggest to use Machine.
Cetin