Introduction Busulfan is one of the most commonly used drugs in preparative regimens for hematopoietic stem-cell transplantation. Administration of accurate doses of busulfan is critical to limit toxicity while ensuring maximum efficacy. Indeed, busulfan pharmacokinetics exhibit important interpatient variability and its therapeutic interval is narrow.Objective To determine the impact of glutathione-S-transferase (GST) genotype and individual characteristics on the pharmacokinetics of oral busulfan in adult patients.Methods An observational retrospective study was performed. Individual characteristics were obtained from the medical charts of adult patients who received oral busulfan between June 2003 and May 2007. Genotype for the enzyme GST A1 was determined in patients for whom consent was available. Linear regression analysis was used to identify the variables that have significant impact on oral clearance of the first dose of busulfan. Oral and intravenous pharmacokinetics were also compared in patients presenting between November 2006 and May 2007.Results Oral clearance on the first dose of busulfan was available for 104 of the 106 eligible patients. Lean body weight (LBW) was used to determine the dose and was a significant predictor of oral clearance (r = 0,61; r2 = 0,37). No added precision was obtained with the addition of age, bilirubin and hepatic transaminase levels (r = 0,62; r2 = 0,38). GST genotype was determined for 87 patients and an increase in precision is obtained when adding GST genotype to LBW for prediction of oral clearance (r = 0,65; r2 = 0,43). The impact of GST genotype was found to be independent of all other variables considered. The model that includes LBW, GST genotype, age, bilirubin and hepatic transaminase levels provides little further precision (r = 0,67; r2 = 0,45). No significant impact was identified when considering the following variables: gender, creatinine clearance, albumin, lactate deshydrogenase and alkaline phosphatase levels, prior history of chemotherapy and/or radiotherapy, concomitant use of acetaminophen or antifungal drugs. The impact of GST genotype (*A wild-type allele, *B variant allele with reduced activity) on oral clearance was also evaluated. A significant difference in busulfan clearance between individuals with the *A/*A genotype and individuals with the *B/*B genotype was observed (p = 0,003; ANOVA and Bonferroni). In the prospective study, eight patients agreed to receive their second dose of busulfan via the intravenous route. Oral clearance was a good estimation of absolute clearance (Student's t-test for mean difference p = 0,573; r = 0,963). Volume of distribution of busulfan was an excellent predictor of oral clearance for the 8 patients (r = 0,96; r2 = 0,93) and LBW was an adequate predictor of volume of distribution (r = 0,94; r2 = 0,88).Conclusion An increase in precision could be obtained by adjusting the busulfan dose with LBW and GST genotype rather than LBW alone. No other individual patient characteristic was identified that influences significantly the pharmacokinetics of oral busulfan. More than 50% of intersubject variability in busulfan oral pharmacokinetics remains unexplained even after considering all above variables. Estimation of volume of distribution may represent a very innovative and promising avenue to more closely predict busulfan dosage.