Cost–benefit models predict that hiding time in refuge increases with predation risk. Studies of diverse prey confirm this prediction for various single risk factors, but much less is known regarding joint effects of variation in multiple factors. For single risk factors, we predicted that Balearic lizards (Podarcis lilfordi) hide longer after faster and more direct approaches by predators and when predators attack persistently or remain close to the refuge. We predicted that effects of approach speed on hiding time would interact with those of directness of approach and predator proximity to the refuge. We simulated attacking predators by approaching lizards ourselves. Predictions for all single risk factors were confirmed: hiding time was greater after faster and more direct approaches, second than first attacks, and when an investigator stood closer to the refuge. These findings agree with those for other prey that are ecologically and phylogenetically diverse, suggesting that optimality theory is broadly applicable to refuge use. The predictions of interaction of approach speed were confirmed for both directness of approach and predator proximity, but the interaction between approach speed and predator proximity was opposite that predicted. Knowledge of the shapes of curves relating joint risks to hiding time are needed to use optimality theory to predict joint effects of multiple risk factors, but these shapes are unknown. Comparative studies are needed to detect quantitative effects of ecological and phylogenetic differences on hiding time.