In a previous reply I did indicate that there was a BEGIN TRAN on the first connection before the first Insert. And I believe you are right that just closing the recordset might be enough to prevent the second connection from being created. I observered the second connection in Query Analyzer while the app is waiting for a timeout to occur. It is also evident in the profile trace. The 2nd connection has to be outside the transaction of the 1st connection and therefore waiting for the lock to check the foreign key as Stax68 indicated.

It is amazing how many wats bad coding can get you.

Thanks to everyone for their help. : )

It may be a few days before a decision is made but I'll let eveyone know which way we go.

Mark Hottinger