134 patients with acquired aplastic anaemia (AA) were given HALG 15 mg/kg/d for 5 d and methylprednisolone for 1 month, and randomized to receive (n = 69) or not (n = 65) oxymetholone 2 mg/kg/d p.o. daily for 4 months. Early mortality (< 120 d) was comparable in the two arms 12/69 (17%) and 11/65 (17%), and correlated with the severity of the disease (39%, 10% and 6% respectively in patients with neutrophil counts (PMN) < 0.2, 0.2-0.5, > 0.5 x 10(9)/l). The response rate at 120 d was significantly greater in patients receiving androgens (56% v 40%; P = 0.04); it was 68% v 48% (P = 0.02) in patients surviving 120 d, and 78% v 27% (P = 0.03) in females with PMN less than 0.5 x 10(9)/l. In a multivariate Cox analysis on patients with less than 0.5 x 10(9)/l PMN, the probability of responding without androgens was reduced compared to the androgen treatment arm (P = 0.05). Survival was comparable in the two groups (71% v 65%). It was superior (74% v 50%), but not significantly (P = 0.1) in females with PMN < or = 0.5 x 10(9)/l receiving androgens. Side-effects, including biochemical abnormalities and virilization, could be controlled and were reversible. In conclusion, the addition of androgens to HALG and methylprednisolone as first line treatment of aplastic anaemia significantly improves the response rate at 4 months, particularly in females with low neutrophil counts, although there is no significant effect on short-term survival. The reversible side-effects warrant the use of androgens as an adjunct to the first course of ALG in females with severe AA.