Prey must balance gains from activities such as foraging and social behavior with predation risk. Optimal escape theory has been successful in predicting escape behavior of prey under a range of risk and cost factors. The optimal approach distance, the distance from the predator at which prey should begin to flee, occurs when risk equals cost. Optimal escape theory predicts that for a fixed cost, the approach distance increases as risk increases. It makes no predictions about approach distance for prey in refuges that provide only partial protection or about escape variables other than approach distance, such as the likelihood of stopping before entering refuge and escape speed. By experimentally simulating a predator approaching keeled earless lizards, Holbrookia propinqua, the predictions of optimal escape theory for two risk factors, predator approach speed and directness of approach were tested. In addition, predictions that the likelihood of fleeing into refuge without stopping and the speed of escape runs increase with risk, in this case predator approach speed, and that lizards in incompletely protective refuges permit closer approach than lizards not in refuges were also tested. Approach distance increased with predator approach speed and directness of approach, confirming predictions of optimal escape theory. Lizards were more likely to enter refuge and ran faster when approached rapidly, verifying that predation risk affects escape decisions by the lizards for escape variables not included in optimal escape theory. They allowed closer approach when in incompletely protective refuges than when in the open, confirming the prediction that risk affects escape decisions while in refuge. Optimal escape theory has been highly successful, but testing it has led to relative neglect of important aspects of escape other than approach distance.