Cindy,

I'm agreeing with you. It turns out creating the index tacks on an additional 2 seconds on average. I had assumed that when it was creating a temporary index, it was a bad thing. In this case, it is better to have SQL create the index than for me to do it manually.

The testing is valid, though. This is how it will operate on a web site. After the first hit, I will not be restarting the machine. I did make sure to close all, clear all to retrieve whatever memory I could retrieve.

>Dan,
>
>How are you testing? VFP caches index data between runs. The only real test is to turn the machine off between tests. Also, eliminate network issues by testing with data on a your hard drive.
>
>
>>Cindy,
>>
>>That's very help info.
>>
>>It shows that it has to create a temporary index even though it is fully optimized. I thought that this was a problem.
>>
>>But in one test I got times from an average of 13 seconds when not creating the index to 15 seconds when creating the index.
>>