Group-size effects, whereby antipredator vigilance decreases as group size increases, are widely reported in mammals and birds but a meta-analysis has only been conducted in birds. We systematically reviewed the literature on mammalian group-size effects, estimated the effect sizes in each study, and conducted a phylogenetic meta-analysis. We obtained 296 effect sizes from 97 species belonging to 10 Orders and 26 Families. Overall, effect sizes indicated a moderate negative effect of group size (r = - 0.44), but 43% of the effect sizes were compatible with a null effect of group size. There was significant heterogeneity in effect sizes. Weaker effect sizes occurred when vigilance was measured as a frequency or a duration rather than as a percentage of time spent vigilant, when measured in closed habitats, during the reproductive season, and in mixed-sex groups or during times when juveniles were absent. We infer a ‘file drawer problem’ because there were relatively few studies with smaller sample sizes reporting small group-size effects. The results confirm the importance of group size in explaining variation in mammalian vigilance, but also suggests that a substantial amount of variation remains unexplained. We suggest that future studies should aim to study mammalian group-size effects by quantifying the percentage of time allocated to vigilance rather than apparently lower-power methods such as frequency or duration of vigilance.