hoo-t's reply is the best way to do it for a result set. 

However, If you want it as a single continuous result, then try this:-

--create test table and populate...

Or something like this :- Which does away with the extra join and the group by

SELECT p.EmployeeID, p.LastName, p.FirstName, p.UpdDate

FROM Personnel p

where p.UpdDate=(Select max(Upddate) from Personnel x where p.EmployeeID = x.EmployeeID)

The order by forces SQL Server to create a temptable. If your column NEWID is unique and incremental, why don't you generate a random number within it's range,pass it...

One other way to write this sort of query - which I find to be a very useful technique is :-

CREATE PROCEDURE au_info @lastname varchar(40), @firstname varchar(20) AS

SELECT au_lname,...

Why don't you compile the Sql and then exec it into a temp table - do your cursor select against the temp table - run the cursor. Then truncate...

It depends on the purpose of your SQL, but, why don't you consider writing a View that unions all the data from the similarly structured tables - you can then...

A bit of a bodge, I know, but you could do :-

declare @header varchar(50)

set @header = 'TestColumn'

select @header

union

select convert(varchar(10),1)

order by 1 desc

It would mean that you'd have to do converts...