The behavioural effects of varying the replacement dose of the oral androgen testosterone undecanoate (TU, Restandol) were investigated in eight hypogonadal men. Each man was withdrawn from his previous androgen replacement regimen and after a period of no treatment was administered four doses of TU (40, 80, 120 and 160 mg/d) in four successive one-month treatment periods using a double-blind design. Throughout the study each man recorded his sexual interest, behaviour and mood state using a daily diary form. In addition, four hypogonadal subjects had sleep erections and sleep quality assessed in the laboratory during a state of androgen withdrawal and again after a minimum of 5 months of androgen treatment. A dose response relationship was demonstrated for frequency of sexual thoughts, arousal accompanying sexual thoughts and well-being. Sleep erections improved significantly during testosterone administration. In men, sexual interest and sleep erections are androgen-dependent.