To determine whether allogeneic bone marrow transplantation (BMT) is associated with a graft vs leukemia (GVL) effect in man, the relapse rate of acute leukemia after allogeneic BMT was compared with that occurring after syngeneic (genetically identical twin) BMT. The patients had ALL or ANL in second or subsequent complete remission (CR) or in relapse. The allogeneic and syngeneic marrow recipients were comparable in diagnosis, age, and interval from diagnosis to BMT and received comparable chemoradiotherapy regimens. Allogeneic marrow recipients, however, received, in addition, GVH disease prophylaxis, most often methotrexate and, more recently, cyclosporine or both. All patients treated by the Seattle team from 1970-1986 are included. Leukemic recurrence was observed in 62% of 785 allogeneic recipients and 75% of 53 syngeneic recipients (p less than 0.0001). The results confirm the circumstantial evidence for a GVL effect exerted by allogeneic marrow. Analyses are in progress to determine whether a GVL effect exists in subsets of patients as a function of their particular diagnosis or status at the time of BMT.