The role of splenectomy in aplastic anaemia (AA) is controversial. The hazards of operating on a severely pancytopenic patient, the fear of compromising the patient's immune function, and the improvement of non-surgical treatment have made splenectomy unpopular in this disease. We have evaluated positive and adverse effects of splenectomy in 80 patients with severe aplastic anaemia (SAA) treated with antilymphocyte globulin (ALG) (group A), using 52 nonsplenectomized ALG patients as controls (group B). All patients survived the operation. Nonfatal complications of surgery occurred in 10 (12.5%). Splenectomy induced a significant increase of peripheral blood neutrophils, reticulocytes and platelets within 2 weeks, followed by a continuous increase of all values over the following weeks. 28/132 patients (21%) developed a late clonal disorder of haemopoiesis, paroxysmal nocturnal haemoglobinuria (PNH) or myelodysplastic syndrome (MDS), or both. Their incidence was identical in groups A and B. 13/28 (59%) died, 10/17 (59%) in group A and 3/11 (27%) in group B (not significant (n.s.)). Overall probability of survival at 18 years after ALG was 51+/-6% for group A and 61+/-7% for group B (n.s.). We conclude that splenectomy in AA is safe. It induces an immediate increase of peripheral blood counts and, thereafter, a continuous improvement of haemopoiesis. It does not increase the incidence of late clonal complications but has a borderline effect on mortality from these disorders. Splenectomy should be reconsidered in selective nontransplanted patients who have prolonged transfusion requirements despite otherwise optimal treatment.