Bhuvnesh
i dont think , it only depends if the any one Non clustrered indexe has all columns from SELECT part plus where clause , other wise there could be key or RID lookup
I think that regardless if the index has the additional columns included or not (using INCLUDE or otherwise), additional columns will generate additional data that has to be piped through one way or another - it doesn't really matter if that additional data comes from the index or from the source table.
Again, the rule really is: the less columns, the better - and I think (but can't confirm just now) that it doesn't matter if these columns are referenced in the WHERE clause or the SELECT statement in terms of performance.
B