One hundred nine patients with hematologic malignancies, undergoing bone marrow transplants (BMT) from unrelated donors, were randomized in 2 consecutive trials to receive or not to receive antithymocyte globulin (ATG) in the conditioning regimen, as follows: (A) 54 patients (median age, 28 years; 39% with advanced disease) were randomized to no ATG (n = 25) versus 7.5 mg/kg rabbit ATG (Thymoglobulin; Sangstat, Lyon, France) (n = 29); (B) 55 patients (median age, 31 years, 71% with advanced disease) were randomized to no ATG (n = 28) versus 15 mg/kg rabbit ATG (n = 27). Grade III-IV graft-versus-host disease (GVHD) was diagnosed in 36% versus 41% (P =.8) in the first and in 50% versus 11% (P =.001) in the second trial. Transplant-related mortality (TRM), relapse, and actuarial 3-year survival rates were comparable in both trials. In fact, despite the reduction of GVHD in the second trial, a higher risk for lethal infections (30% vs 7%; P =.02) was seen in the arm given 15 mg/kg ATG. Extensive chronic GVHD developed overall more frequently in patients given no ATG (62% vs 39%; P =.04), as confirmed by multivariate analysis (P =.03). Time to 50 x 10(9)/L platelets was comparable in the first trial (21 vs 24 days; P =.3) and delayed in the ATG arm in the second trial (23 vs 38 days; P =.02). These trials suggest that (1) 15 mg/kg ATG before BMT significantly reduces the risk for grade III-IV acute GVHD, (2) this does not translate to a reduction in TRM because of the increased risk for infections, and (3) though survival is unchanged, extensive chronic GVHD is significantly reduced in patients receiving ATG.