After failure to confirm an anabolic action of testosterone and its derivatives in rats, methandienone ('Dianabol', an "anabolic steroid" used by athletes) has been given to 11 athletic men during a course of weight-training, in a double-blind, crossover experiment. The dose of methandienone was 100 mg/day for 6 wk. Body weight and composition, muscular strength and performance, and indices of endocrine function were studied. Compared with the placebo period, on methandienone the subjects gained weight (mean 3-3 kg +/- 0-6 kg) and accumulated a disproportionately large amount of potassium (420+/-68 mmol); the increase in weight was confined to the lean part of the body, and the muscles increased in size. Strength and performance improved over each training period, but not significantly differently on drug and placebo. On the drug, plasma-cortisol concentration and urinary cortisol excretion increased, and plasma-testosterone decreased. Although the weight and body-composition changes may demonstrate an anabolic action of methandienone in man, they may alternatively have been caused by an increase in intracellular fluid, and the question of anabolic action therefore remains open.