High-ranking male and female vervet monkeys in Amboseli National Park alarm-call at higher frequencies than do low-ranking males and females. This correlation between rank and alarm-calling does not occur because high-ranking animals have more offspring or close kin than low-ranking animals. Similarly, high-ranking animals do not seem to have greater opportunities to spot potential predators; they do not scan the habitat more than low-ranking animals, nor are they more likely to be in the forefront of group progressions. It seems possible that low-ranking animals may spot predators as often as high-ranking animals, but that they do not always alert other group members to approaching danger. As a further test of the possibility that monkeys may vary the rate of alarm-calling depending on social contexts, an experiment was conducted on captive vervet monkeys in which adult females were presented with a predator in the presence of either their offspring or an unrelated juvenile. Adult females alarm-called significantly more often when with their offspring than with unrelated juveniles. A similar experiment conducted on adult males showed that males alarm-called at higher rates in the presence of adult females than in the presence of other adult males. The results of both experiments and observations suggest that monkeys may vary the rate at which they warn others of danger. The withholding of information may be an effective means to deceive others, because it is relatively difficult to detect cheaters. The possible advantages of selectivity in alarm call production are discussed.