:: My 10 second experiment shows both sections of the union
:: query using the best index in each case. Possibly more
:: efficient if both datasets contain duplicates, as only one
:: of each record will be sent across the network. On the
:: downside, you need to wait for both queries to be run before
:: you get the information back.

Thanks. Im not doing a distinct between the unions, it just makes processing
easier in our case to have one result instead of two. Network traffic is not
an issue either, its done server side.

:: I would only use it if the fields matched up between the two queries.

They do.