Thanks Christof, but the KB article raises far more questions than it answers (typical for anything written by Microsoft.) It is logical that all connections would be terminated by the same period of inactivity and therefore none would remain active. How would pooling help in that instance? How much of an application needs to be rewritten to use connection pooling? Does it alter the look-and-feel fo the application? What happens to the SPT processes that use the statement handle? Lots of other questions come to mind but that will get me started...