Predation is considered one of the most important selective pressures on free-ranging animals. Our understanding of it derives mainly from studies of individual vigilance (visual scanning of the surroundings beyond the immediate vicinity) and aggregation in prey. Vigilance bears a direct relationship to aggregation, because animals in groups may rely on associates for early warning of danger. This review addresses the relationship between vigilance and aggregation with particular attention to the prediction that individual vigilance declines with increasing group size. Contrary to most other animals studied, primates do not support the prediction. Exploring this, I examined the assumptions underlying vigilance theory in the light of primate behaviour. First I tested whether manual harvesting and upright processing of food as seen among primates might permit them to feed and scan simultaneously. I found no support for this idea. Next I examined the targets of primate vigilance and found that one component (within-group vigilance) might explain the differences between primates and other animals. Finally, I evaluated whether individual primates in large groups face a lower risk of predation than those in small groups. A conclusion was impossible, but by separating group-level from individual-level risk, I was able to identify several common circumstances in which group size would not predict individual risk or vigilance. These circumstances arose for primates and nonprimates alike. I concluded that the relationship of vigilance to aggregation is not straightforward. The absence of a group-size effect on vigilance among primates is probably due to functional differences in vigilance behaviour or safety in groups, not to methodological differences. Furthermore, future work on animal vigilance and aggregation must fully consider both the targets of glances, and the assumption that larger groups are safer from predators. I predict that animals will not relax vigilance in larger groups if conspecific threat increases with group size. Group size will not predict individual risk of predation nor individual vigilance rates when predators do not rely on surprise, or when predators select a small subset of highly vulnerable group members. Copyright 2000 The Association for the Study of Animal Behaviour.