Do predator–prey ‘waiting games’ where prey hide from potential predators have inherently unstable evolutionary outcomes, making it impossible to generate quantitative predictions about hiding times? Fiddler crabs, Uca lactea perplexa, respond to potential predators by retreating into their burrows. Time inside the burrow during unprovoked retreats during normal activity provides a ‘null model’ to test whether sex, tidal cycle and body size affect hiding time from potential predators. Using experimentally created predator-like stimuli we found that males hid for significantly longer than females, and larger crabs of both sexes also hid for longer. This differs from burrow use during unprovoked retreats, suggesting hiding time varies depending on the potential risk of predation on re-emergence. If risk prior to hiding predicts risk on emergence, the closer the proximity of a predator-like stimulus when first encountered the longer crabs should hide. We confirmed this experimentally (stimuli at 0.5 versus 2.5m). Finally, we tested whether males hide for longer when a predator-like stimulus approaches them directly rather than tangentially. None of three pairwise comparisons was statistically significant, but crabs hid less as the angle of approach became more tangential. These results suggest prey can use stimuli prior to hiding to predict predation risk on re-emergence, but studies on predators are required to test this claim. Finally, theoretical models must explain why hiding time has a lognormal distribution and low variance such that a predator can predict when most prey will re-emerge. For example, 95% of crabs re-emerged within 2.3min of hiding. Copyright 2003 Published by Elsevier Science Ltd on behalf of The Association for the Study of Animal Behaviour.