Nadya,
How many matching records are there in btcZipcode? With other words, how many records do you have in the result set? If you have only a few zipCodes in btcZipcode, you could build a where clause using IN() or INLIST() instead of joining the two tables. For example if btcZipcode contains the zipCodes "A", "B", "F" and "X", you could run the following query:

SELECT sitemstr.* ;
       from o:\redp\dbc\ma\sitemstr ;
 WHERE sitemstr.zipcode IN ("A", "B", "F" and "X") ;
INTO TABLE ..\..\work\query\qryU0ZA34J

I have used this technique to increase performance, but it only works if you have less than 25 elements.

Another thing to try is to SET DELETED OFF and see if this makes any difference.

HTH
>Hi everybody,
>
>I'm running this simple test:
>
>

>SELECT sitemstr.* ;
>       from o:\redp\dbc\ma\sitemstr ;
>       inner join c:\wistemp\btc_0IU0ZA4P5 btcZipcode on sitemstr.zipcode = btcZipCode.btcCode ;
>        INTO TABLE ..\..\work\query\qryU0ZA34J

>
>Temp table has an index on btcCode and SiteMstr has an index on ZipCode. sys(3055,11) shows, that optimization level for SiteMstr is full and for temp table is none. The query executes very long time (SiteMstr has ~3mln. records). The original query (not the test one) has 7 joins already and it takes forever. Is it here a way to optimize it?
>
>I've tried to change select ... where ZipCode in (select btcCode from ) form and it also shows none optimization for temp table, though seems to execute a little bit faster (still very slow).