>I've lent out my ADO book, but that behavior makes sense because a command object is simply executing a command, with no back link to the data source.

I don't quite follow you here. Both methods return an ADO clientside recordset that has a declared locktype of adLockBatchOptimistic. Are you saying that without a back link to the data source, the locktype has to be coerced to adReadOnly? Can you explain why that would be?

The reason I was using a Command object is to get the better performance available thru a SP. Seems a little funny for a limitation like this to be imposed.