Selecting the optimal subset is highly beneficial to numerous developments in simulation optimization. This paper studies the problem ofmaximizing the probability of correctly selecting the top-m designs out of k designs under a computing budget constraint. We develop a new procedure which is more efficient and robust than currently existing procedures in the literature. We also provide an analysis on its asymptotic convergence rate. Based on this analysis, we show that our new procedure achieves a higher convergence rate than other procedures under certain conditions. Numerical testing supports our analytical analysis and shows that the new procedure is significantly more efficient and robust.