This talk presents a method for finding multiple, high-quality minima of simulation-based functions that require significant computational effort to evaluate. We first introduce select methods for derivative-free optimization, and then present a multistart algorithm we have recently developed that utilizes concurrent evaluations of the simulation in order to more efficiently search the domain. Theoretical properties of the framework are presented as well as results from extensive numerical experiments. We show that the method's ability to find multiple minima for a set of benchmark problems scales well with the number of concurrent evaluations.