> Dear Adam,
>
> Thanks for trying it in your machine. How about the RowsAffected
> statistics in your case?

Unfortunately I don't have any admin tool that reports a statistic
called this.

>
> I'm using FB2.0.1.

I don't have time today to upgrade my laptop to test this problem,
perhaps you could try 1.5 to see if this is a new issue in the 2.x
optimiser.

> This is much better than 28 seconds but it is
> still much higher because I expect it to be sub-second if it can
> choose the right index.

It certainly should be sub-second.

If it is indeed a 2.x issue, you could probably write a .sql script to
recreate an example database (using a stored procedure to populate
data). Post it back here to see whether anyone can fault it and if not
it should be submitted as a bug report.

Adam