Theory and empirical evidence suggest that predator activity makes prey more wary and less vulnerable to predation. However if at least some prey in the population are energetically or spatially constrained, then predators may eventually increase local prey vulnerability because of the cumulative costs of anti-predation behaviour. We tested whether repeated attacks by a predator might increase prey vulnerability in a system where redshanks on a saltmarsh are attacked regularly by sparrowhawks from adjacent woodland. Cumulative attack number led to a reduction in redshank numbers and flock size (but had no effect on how close redshanks fed to predator-concealing cover) because some redshanks moved to safer but less profitable habitats, leaving smaller flocks on the saltmarsh. This effect held even though numbers of redshank on the saltmarsh increased with time of day. As a result of the change in flock size, predicted attack-success increased up to 1.6-fold for the sparrowhawk, while individual risk of capture for the redshank increased up to 4.5-fold among those individuals remaining on the saltmarsh. The effect did not arise simply because hawks were more likely to attack smaller flocks because attack rate was not dependent on flock size or abundance. Our data demonstrate that when some individual prey are constrained in their ability to feed on alternative, safer foraging sites, their vulnerability to predation increases as predator attacks accumulate, although those, presumably better quality individuals that leave the immediate risky area will have lower vulnerability, so that the mean vulnerability across the entire population may not have changed substantially. This suggests that the selective benefits of multiple low-cost attacks by predators on prey could potentially lead to 1) locally heightened trait-mediated interactions, 2) locally reduced interference among competing predators, and 3) the evolution of active prey manipulation by predators.