For cold start users, trust-aware recommender systems have been shown to be more useful in providing better quality recommendations than traditional systems. By exploiting the concept of trust propagation over the trust network, trust-aware systems are able to make large number of accurate recommendations by utilizing the few trust statements given by the new user. In this paper, we propose intelligent strategies based on efficient heuristics, and examine existing strategies on the basis of which prospective trusted users are selected to be presented before a cold start user. Through experimental evaluation we show that selecting prospective trusted users based on our proposed efficient heuristics i.e. hybrid parameter preference score gives the most accurate recommendations. We also propose a clustering approach to select n users for the final list from the set of prospective users selected by a strategy. Furthermore, our paper also describes a new metric entropy0 to reduce the effort required by a cold start user to evaluate prospective trusted users.