Selection for divergent performance optima has been proposed as a central mechanism underlying adaptive radiation. Uncovering multiple optima requires identifying forms associated with different adaptive zones and linking those forms to performance. However, testing and modeling the performance of complex morphologies like the cranium is challenging. We introduce a three-dimensional finite-element (FE) model of the cranium that can be morphed into different shapes by varying simple parameters to investigate the relationship between two engineering-based measures of performance, mechanical advantage and von Mises stress, and four divergent adaptive zones occupied by New World Leaf-nosed bats. To investigate these relationships, we tested the fit of Brownian motion and Ornstein–Uhlenbeck models of evolution in mechanical advantage and von Mises stress using dated multilocus phylogenies. The analyses revealed three performance optima for mechanical advantage among species from three adaptive zones: bats that eat nectar; generalized insectivores, omnivores and some frugivores; and bats that specialize on hard canopy fruits. Only two optima, one corresponding to nectar feeding, were consistently uncovered for von Mises stress. These results suggest that mechanical advantage played a larger role than von Mises stress in the radiation of New World Leaf-nosed bats into divergent adaptive zones.