>>
>>It's been that way since MDAC 2.1. And as John states, with a few tweaks it can run even faster.
>>
>
>What? That is not what John said. He said that the results are not meaningful. Using client-side recodsets *<g> will make the results much slower, and will be a more fair comparison to SPT.

You're right, John was saying something else.

What I was trying to point out is that since MDAC 2.1 ADO is faster than SPT regardless of cursor type, even with the use of the same ODBC connection as opposed to the more direct OLEDB (when available).