This paper presents an experimental evaluation of evolutionary pattern search algorithms (EPSAs). This analysis provides a first step towards understanding of how the broad theory of EPSAs needs to be applied to define practical EPSAs. We evaluate the effect that different types of mutation operators have on the frequency with which an EPSA could update its step length, as well as the effects of different crossover rates. Additionally, we describe a stopping rule for EPSAs that empirically terminates them near a stationary point. The empirical performance of EPSAs confirms that they can perform nonlocal optimization on standard test optimization problems.