Perhaps my experience is unusual, but I've recently determined that clustered indexes do not perform nearly as well as non-clustered indexes. Perhaps this is because the tables I was working with were very wide, but non-clustered indexes which were covering significantly outperformed the clustered index (this is on SQL Server 2005). I now keep my eye out for cases where there is a clustered PK and consider creating a non-clustered version and add any columns not in the key which are typically needed to be included in the index.