Walter,

Sorry, I wasn't reading the point of your message correctly last night.

Yes, the join would cause multiple rows in the result if the child table had multiple rows, adding DISTINCT would prevent that from occuring.

EXISTS does guarantee 1 or 0 row in the result per pk.

>Where both can return different results depending on the data.
>Only the latter ensures that records of table1 occur only zero or ones in the resultset.