This study examines vigilance as a behavioural indicator of the importance of infanticide risk by comparing the infanticide avoidance hypothesis with the predation avoidance and mate defence hypotheses for wild Thomas's langurs (Presbytis thomasi) in Sumatra. We found that all individuals were more vigilant in situations of high predation risk, i.e. lower in the trees and in the absence of neighbours. Females were also more vigilant on the periphery of the group. However, there were variations in vigilance levels that could not be accounted for by the predation avoidance hypothesis. Males without infants showed higher levels of vigilance in areas of home range overlap than in non-overlap areas during the early phase of their tenure, strongly suggesting mate defence. In these areas of home range overlap where Thomas's langur groups can interact, males may attack females and infants, and so the infanticide risk for males and females with infants is likely to be high in these areas. Only females with infants, but not males with infants or females without infants, showed higher vigilance levels in overlap areas than in non-overlap areas; in addition, in overlap areas, females with an infant were more vigilant than females without an infant, while this was not the case in non-overlap areas. Both females and males with infants were more vigilant high in the trees than at medium heights in overlap areas but not elsewhere. These ®ndings can only be explained by the infanticide avoidance hypothesis. In contrast to predator attacks, infanticidal male attacks come from high in the canopy, and only occur in overlap areas. There was a signi®cant sex dierence in vigilance, but males were only more vigilant than females without an infant, and not more vigilant than females with an infant. We conclude that vigilance varied mainly in relation to the risk of predation and infanticide. Mate competition only played a role for males during the early phase of their tenure. Predation risk seems to oer the best explanation for vigilance for all individuals in the absence of infants. Both predation risk and infanticide risk played a role for females and males with infants.