Jean-Rene:

I wrap everything up in a transaction and if something fails, I roll it back.

If you want to limit your result sets, you send a parameter to the stored procedure, like a primary key for example. Same goes for updates and deletes. For inserts, I don't send any parameters.

Here is a simple example of a SQL Server stored procedure that receives as a parameter a primay key that identifies the record to be retrieved:

CREATE procedure spGetXrefs
@caseid int
as
select * from Xrefs
where fk_cases = @caseid

Charlie