In a single-centre study the feasibility and efficacy of repeated antilymphocyte globulin (ALG) for patients with severe aplastic anaemia (SAA) not responding to an initial ALG treatment or relapsing after initial response to ALG was evaluated. 139 consecutive patients with newly diagnosed SAA were treated with ALG between 1976 and 1995. 89 patients responded to a first course; 50 patients did not become transfusion independent. Of the 89 responders, 66 remained in remission, 23 relapsed. 43 patients received a second or subsequent course of ALG for failure to respond (n = 25) or relapse (n = 18) and were given a total of 53 courses. Acute reactions in the multiply exposed patients occurred during the first ALG treatment in 11 (26%) and during subsequent exposures in 16/53 courses (30%; P > 0.2). Incidence of serum sickness was 63% (27/43) after the initial course compared to 57% (30/53) after subsequent courses (P > 0.2), but clinical signs of serum sickness occurred earlier after repeated (median 6 d) as compared to initial exposure (13d; P = 0.008). Transfusion-independent haemopoiesis was achieved in 27/43 (63%) and survival probabilities for the 43 patients receiving multiple courses of ALG was 52 +/- 8% at 10 years. The probability of developing a late clonal disorder was 53 +/- 10% after multiple, as compared to 34 +/- 7% after single exposure (P = 0.15). No difference in results was observed between patients retreated for failure to first ALG or for relapse. ALG of the same species can be repeated without increased risks of side-effects in patients with SAA. A second or subsequent course of ALG from the same source can be effective when the first course has failed.