I have a very simple update command that is taking an enormous amount of time to run in a production installation. In test runs with about a million and a half rows it runs in about 40 seconds, but in the installation with I would guess around 10 million rows, it runs in 11 hours! The query is shown below: UPDATE IMP_HDR SET CUST_ID = C.CUST_ID, CUST_MATCH_CODE = 'C' FROM CUST C WHERE C.CLIENT_ID = IMP_HDR.CLIENT_ID AND C.LAST_CUSTOMER_ID = IMP_HDR.CUSTOMER_ID ; The primary keys IMP_HDR.ID (not involved in query) and CUST.CUST_ID They are both Non-clustered. Indexes also exist on IMP_HDR.CUST_ID, CUST.(CLIENT_ID,LAST_CUSTOMER_ID) Simple recovery model logging is being used. My best guess is the changing of the data in the IMP_HDR.CUST_ID field from null to a value is expanding the record size causing all pages to be rewritten. Is that possible? Any other opinions, or insight on how to fix this is greatly appreciated. Thanks, Dale Lear.