which index is best

If the table does not have a primary key, it does not necessarily mean that it does not have a clustered index (primary key can be enforced by a unique non-clustered index). While it is legal to have heap tables (the ones which don't have a clustered index), it is definitely not a good practice (though there are rare scenarios when the heap tables are justified). However, just the fact that there is a column in the table which is "often used by 'where' condition", it does not yet mean that such column is a good candicate for a clustered index. There are 2 bare minimum requirements which should be met to satisfy the clustered index candidate criteria:

• The values should be ever increasing (every new record inserted into a table must have the clustered column(s) value(s) greater than those in existing records)

• The column(s) should be as narrow as possible. This is because clustered index columns's values become a part of of the leaf pages of every other existing nonclustered index.

And finally, a table can have only one clustered index.

The best introduction to indexes I ever read were the 3 part series by Gail Shaw, beginning from this one: http://www.sqlservercentral.com/articles/Indexing/68439/ Please read them as they are enormously helpful.

Oleg