Just caught up with your response to my question about corrupt indexes posted almost a year ago. Sorry I missed this. I appreciate your input. For what it's worth, we seem to have solved the problem by issuing FLUSH after the first INSERT command. Thanks again for your help.