Like many other non-human primates (Lancaster & Lee, 1965; Michael & Zumpe, 1971), rhesus monkeys in the wild (Vandenbergh & Vessey, 1968) and in captivity (Michael & Keverne, 1971; Michael, Zumpe, Plant & Evans, 1975; Robinson, Scheffler, Eisele & Goy, 1975) exhibit seasonal changes in sexual behaviour. In the northern hemisphere there is a peak in ejaculatory activity in the late autumn and early winter followed by a decline in late winter and spring when, in the wild, most fertile females are pregnant. However, pregnancy is not the sole cause of the decline in sexual activity. Pairs of intact monkeys in which the females were sterilized by ligation of the oviducts (Michael & Keverne, 1971) and pairs in which the females were ovariectomized and treated s.c. with 10 μg oestradiol benzoate/day (Michael et al., 1975) showed the seasonal changes characteristic of intact animals in the wild. It seemed, therefore, that the annual behavioural rhythm might be in part dependent on changes within the male of the pair. In two studies (Plant, Zumpe, Sauls & Michael, 1974; Robinson et al., 1975), each of which included 1 complete year, well-marked seasonal changes occurred in the plasma testosterone levels of male rhesus monkeys, with maxima in the autumn and winter months. A notable feature of our preliminary results was the occurrence of an autumnal increase although animals were maintained in a constant photoperiod. These studies have now been extended for a 3-year period during which environmental conditions were rigorously controlled to assess the extent to which long-term rhythms in plasma androgen levels are dependent on known exteroceptive factors.