Hi,

Have you tried running the queries in a different order (first the NOLOCK variant and then the normal one)?  And running them with MAXDOP 1?

Anyway, there is a little catch with NOLOCK however.  SQL Server mostly uses unordered clustered index scans when NOLOCK is specified, this can cause duplicate records to be returned when page splits occur while your query is scanning the index.

Kind regards,

Wesley