This experiment was designed to elucidate the role of testosterone in the decline of sexual behavior in aging males. Old (25 months) and middle-aged (10 months) male rats were given six tests (30-min long) of sexual behavior. The old males then were divided into two groups: intact and castrated with testosterone treatment. The middle-aged males were divided into three groups: intact, castrated with testosterone present all the time, and castrated with testosterone present only when tested. The old males were given another set of six tests 1 week after the operation (when 27.5 months old), and the middle-aged males were given three more sets of six tests 8, 26, and 39 weeks after the operation (when 15, 19, and 22.5 months old, respectively). Blood drawn after each test set was assayed for testosterone. The middle-aged intact males had higher levels of testosterone than the old intact males, and the testosterone-treated castrated males, whether middle-aged or old, had higher levels than the intact males. The presence of higher levels at the time of testing resulted in increased rates of mounting and intromitting in old males and an attenuated decline in the mount rate and percentage of tests with intromissions in middle-aged males. The middle-aged castrated males with continuous testosterone differed from the middle-aged castrated males with periodic testosterone only in the mount-intromission interval; the former group had a longer interval. It was concluded that testosterone, in general, does not prevent or reverse the decline in sexual performance of aging male rats and that the degree and rate of decline do not depend on whether or not testosterone is continuously present.