Fifty-four patients representing all the cases of acute myeloid leukemia in patients aged over 60 years, presenting to three adjacent hospitals, were studied. Only 26 of the 54 patients were considered suitable for remission induction with intensive combination chemotherapy which produced seven complete remissions (CR) (26 per cent). Eighteen of the 54 patients survived longer than three months--11 of these had received remission induction chemotherapy (five CRs), two low dose cytarabine, one vincristine and vitamin A and four supportive treatment alone. By six months all the patients who had received supportive treatment had died. Patients who received intensive chemotherapy spent 77 per cent of the first 90 days in hospital and half died in hospital. Patients receiving differentiating agents or supportive care spent more time at home (37 per cent, 34 per cent of the first 90 days, respectively) but had shorter overall survival. Assessment of clinical characteristics in an attempt to predict response to treatment and survival indicated that poor performance status and the presence of infection were the most important factors. Analysis was limited by the statistically small number in the group. At present there is no reliable method of predicting which patients will respond well to treatment with intensive chemotherapy. The clinical dilemma is whether to treat more intensively to benefit a minority, or to use supportive treatment to allow more time to be spent at home albeit with a shorter overall survival.