The optimum management of children with severe malarial anaemia is still uncertain. Hence, we have undertaken a study to determine whether iron treatment is as effective at restoring haemoglobin levels one month after presentation as blood transfusion without iron treatment in children with moderately severe malarial anaemia. Two hundred and eighty-seven children with a packed cell volume (PCV) < 15% and malaria infection were recruited into the study; 173 children were assigned to receive blood transfusion because they had a PCV < 12% and/or signs of respiratory distress and the remaining 114 children were allocated at random to receive either blood transfusion (58) or treatment with oral iron (56) for 28 d. Twenty-four children died, 23 in the most severely anaemic group. Fifteen children (65%) died before transfusion was given and most deaths occurred within the first 4 h of admission. One child died in the iron treatment group and 10 subsequently required transfusion. Among the severely anaemic children, those with respiratory distress were at greater risk of death than those without respiratory distress. After 28 d, haematological restoration was significantly better in children who had received iron than in those treated by blood transfusion (P = 0.02). Children who received malaria chemoprophylaxis after discharge from hospital had fewer episodes of malaria and subsequent admissions to a hospital or health centre than those who did not. Children with severe anaemia and clinical signs of respiratory distress must be identified quickly and transfused as soon as possible. However, for less severely anaemic children who are clinically stable, iron therapy offers an alternative to transfusion provided such children can be kept under surveillance and transfused subsequently should this become necessary.