Graft-versus-host disease (GVHD) is a major obstacle to successful bone marrow transplantation (BMT) from matched unrelated donor (MUD). Currently available HLA-A, -B, and -DR serologic testing may not be sensitive enough to detect clinically relevant donor/recipient (D/R) nonidentity. Better HLA matching of D/R pairs using molecular typing for class II antigens in combination with intensive GVHD prophylaxis may potentially reduce the incidence of GVHD and lead to an improved outcome of MUD transplantation. Between July 1991 and August 1993, thirty consecutive patients with hematologic malignancies underwent MUD transplantation from donors who were identical for HLA -A, -B, and -DR by serologic typing. Twenty-five D/R pairs were matched for DRB and DQB by molecular typing (restriction fragment-length polymorphism and sequence-specific oligonucleotide probe hybridization analyses), whereas five were allele mismatched at either DRB or DQB. All patients also received GVHD prophylaxis with the combination of cyclosporine (CSA), methotrexate (MTX), and prednisone (PSE). The median age was 35 years (range, 15 to 50). The diagnoses were: chronic myelogenous leukemia (CML) in chronic phase (CP) (16), CML in more than CP (3), acute leukemia in more than first complete remission (CR) (8), acute leukemia in first CR (1), and advanced high-grade lymphoma (2). The preparative regimen consisted of 1,320 cGy fractionated total body irradiation (FTBI) and 60 mg/kg cyclophosphamide (CY) daily for 2 days in 17 good-risk patients (CML/CP and acute leukemia first CR); and 1,320 cGy FTBI in combination with 60 mg/kg etoposide and 20 to 60 mg/kg CY in 13 patients with advanced leukemia and lymphoma. All patients received CSA, PSE, and MTX on days 1, 3, 6 for GVHD prophylaxis, and 10 patients also received day +11 MTX. All patients engrafted except one who died early of regimen-related toxicity. The incidence of grade III or IV acute GVHD was 24% (95% confidence interval [CI], 10% to 44%) and that of extensive chronic GVHD was 65% (95% CI, 43% to 84%). At a median follow-up of 13.6 months, 57% of the patients are alive in remission with a median Karnofsky performance status of 90%. The cumulative probability of 2-year disease-free survival for all patients was 53% (95%) CI, 33% to 71%); for good-risk patients, 71% (95% CI, 46% to 87%) and for the poor-risk group, 34% (95% CI, 13% to 64%).(ABSTRACT TRUNCATED AT 400 WORDS)