If the large table contains thousands of records and the application is accessing 35% of the table which method to use ?

I think we need more information for a definitive answer. Are you asking what should be done or what will be done. For the former we need to know how many columns there are, how big they are, how much data they contain, how many are being selected, and probably a few other things. For the latter we also need to know what version of Oracle, whether object and system statistics have been gathered, and even then it would still just be an educated guess. Do you have a query doing a full table scan when you think it shouldn't? Can you show the query and provide some information on the tables?