I have tried that. It returns: no rows selected. The difference between two of the databases I have is that on one the query causes 538 physical reads (this is the one that statspack flags). On the other it causes none. Possibly the full table scan if IND$ is getting aged out of the buffer cache, thus causing the reads when the statement is reissued every hour. I still don't know what the source is; I may trace the entire instance for an hour overnight and trap it that way. Thanks Ray _____ From: Jared Still [mailto:jkstill@xxxxxxxxx] Sent: Saturday, January 07, 2006 6:13 PM To: ray@xxxxxxxxxxxxxxxxxxxxx Cc: oracle-l@xxxxxxxxxxxxx Subject: Re: Source of expensive SQL Just a thought: Try running the query yourself and see what the results are. That might give a clue. Jared