Methotrexate has been used as the mainstay therapy to prevent or ameliorate graft-versus-host disease (GVHD) in allogeneic bone marrow transplantation. We began a nonrandomized study in which methotrexate was not given routinely. Fifty-five patients underwent transplant for acute leukemia (44 patients), aplastic anemia (6 patients), and other malignancies (5 patients). Methotrexate was given to 34 patients (MTX +) and was withheld in 21 patients (MTX -). Median (range) age of patients was 12 (0.8-43) years in the MTX + group, and 16 (3-45) years in the MTX- group. Mean days (+/- SEM) to engraftment (neutrophils greater than 500/microL, and platelets greater than 20,000/microL untransfused) occurred earlier in the MTX- patients (19.6 +/- 1.4 v 24.9 +/- 1.8 days for granulocytes, and 19.3 +/- 1.5 v 27.4 +/- 2.8 days for platelets, P less than .05). There were no statistically significant differences between the patient groups for the incidence or severity of GVHD (10/34 in the MTX + group had grade O-l GVHD compared to 9/21 in the MTX- group). The interstitial pneumonitis occurred at a significantly increased rate in patients who received methotrexate (15/34) compared to those patients who did not (3/21) (P = .02). However, there was also a significant relationship between the interstitial pneumonitis and the preparative regimen: if the preparative regimen contained 1,000 rad single fraction total body irradiation, 8/14 patients were affected compared to 5/22 patients affected when 1,200 rad fractionated total body irradiation was used (P = .03). Because methotrexate significantly retards hematopoietic reconstitution, randomized trials for GVHD prevention are recommended.