Hepatic venoocclusive disease (HVOD) is a complication of allogeneic hematopoietic stem cell transplantation (HSCT) and is a well-recognized dose-limiting toxicity of oral busulfan (Bu)-based preparative regimens. The unpredictable absorption of oral Bu from the gastrointestinal (GI) tract and hepatic first-pass effects have led to the development of an intravenous Bu preparation (i.v. Bu). The purpose of this retrospective comparison was to evaluate the incidence rate of HVOD and the 100-day mortality rate in patients treated with a busulfan/cyclophosphamide (BuCy2) regimen in which either oral Bu or i.v. Bu was administered. Data from 2 similar groups of patients treated between March 1995 and December 1997 were analyzed. Thirty patients were treated with oral Bu (1 mg/kg x 16 doses) at City of Hope and 61 patients were treated with i.v. Bu (0.8 mg/kg x 16 doses) in a multicenter trial involving 7 sites. Bu was followed by Cy (60 mg/kg x 2 days) and a histocompatible-sibling-donor HSCT. In the i.v. Bu treatment group, 48% of the patients were classified as heavily pretreated (> or = 3 prior chemotherapy regimens, prior radiation, or prior HSCT) with 13% having had a prior HSCT and 75% having active disease at the time of transplantation. According to the same classification criteria, 33% of the patients in the oral-Bu treatment group were considered heavily pretreated, with 23% having had a prior HSCT and 80% having active disease at the time of transplantation. The incidence rates of clinically diagnosed HVOD were 5/61 (8%) and 10/30 (33%) after i.v. and oral Bu, respectively. HVOD-related mortality occurred in 2 (3.3%) of 61 i.v. and 6 (20%) of 30 oral Bu patients. The (standardized) Jones criteria for HVOD were met by 4.9% of i.v. and 20% of oral Bu patients. Univariate logistic regression analysis identified oral versus i.v. Bu (P = .001) and a diagnosis of myelodysplastic syndrome (P = .04) as statistically significant factors in the development of HVOD, with prior extensive treatment identified as marginally significant (P = .25). No other demographic parameter was found to be significant. After adjustment for prior treatment, multivariate analyses showed that the use of oral versus i.v. Bu was the strongest predictor for development of HVOD (odds ratio, 7.5; 95% confidence interval, 2.1-27.2; P = .002). This study showed that the incidence rate of HVOD is significantly lower (P = .002) and the 100-day survival rate significantly higher (P = .002) in patients treated with i.v. Bu than in patients treated with oral Bu when Bu is used as part of a BuCy2 preparative regimen for allogeneic HSCT.