The probability that lizards would capture crickets declined with distance from the snout, at rates which were significantly more rapid in all directions in Lacerta vivipara than in Podarcis muralis or L. viridis, i.e. the former species responded to potential prey over a smaller area. Capture probabilities at any distance in front of or behind the snout were lower in P. muralis or L. viridis which were pausing during locomotion than in basking lizards, confirming previous results with L. vivipara. Using capture probabilities for pausing lizards to calculate the average time it would take to find a single item of prey (tf) in relation to the mean length of locomotor bursts, on the assumption that prey could only be detected while a lizard was pausing, showed that actual mean burst distance corresponded exactly with the burst distances which gave rise to minimum tf in L. vivipara. Mean locomotor burst distances in P. muralis and L. viridis were lower than the distances which gave minimum tf values. It is suggested that, in these species, the mean burst length has evolved as a compromise between minimising tf and avoiding the high overall energy expenditures which would result from long burst lengths.