I have a small 12 GB database to look after plus I am not a DBA since I am actually one of the developers that developed this database and I know by now how popular developers are amongst DBA's but I base my indexing on the following.

1. A primary key is the clustered index

2. All other indexes are nonclustered since you can anyway only have one clustered index.

The reason why is, as someone has said already, that with clustered indexes the data gets ordered according to the columns and therefore the data gets added that way in the index. Less scanning.

I use identity columns only when absolutely necessary.:-D:-D:-D:-D:-D:-D