High-dose busulfan is widely used in allogeneic and autologous marrow transplantation preparative regimens. Variation in the area under the concentration/time curve (AUC) for oral busulfan results in substantial risk of over or under treatment with excess risk of toxicity or relapse. Use of the IV formulation reduces this variability by eliminating variability in absorption. Variability due to drug metabolism remains, but simplified pharmacokinetic study may be employed to achieve a specific target AUC. In conventional sampling strategies for determining AUC after oral administration, 12 samples are used over 6 hours to assure accurate tracking of erratic absorption. With IV busulfan there is no necessity for measuring plasma levels during the infusion because busulfan pharmacokinetics are well described with a single-compartment, first-order elimination model. In theory, only peak and trough levels should be necessary, but for assurance of reliability in clinical decision making, it must be possible to identify outlier values. This process requires at least 4 samples. We studied a total of 59 adult patients receiving a 2-hour IV busulfan infusion to develop a limited sampling strategy (LSS). At the end of a 2-hour infusion, we collected 11 samples from 18 patients and compared the AUC obtained when all samples were used with the AUC obtained when samples were collected only hourly. The mean AUC calculation was 5% higher (1002 versus 956 microM-min) and the coefficient of variation (CV) was substantially better (4.6% versus 8.2%) when only the postinfusion samples were used. A follow-up study of 41 consecutive patients demonstrated that all patients were easily evaluable with a coefficient of variation (CV) for the AUC of 2.6%. To validate this approach, we analyzed pharmacokinetic data on 60 patients in the phase II clinical trial of the IV formulation described by Anderson et al. Data on an additional 36 patients from a companion study also were analyzed. The AUC based on all 11 samples from each patient were compared with the AUC based on the 5 postinfusion samples. The results of this analysis confirmed comparable reliability and possibly superior precision of the University of Alabama at Birmingham 5-sample LSS. These results validated that LSS for IV busulfan will make possible meaningful and accurate comparisons of busulfan versus TBI-based preparative regimens and comparison of dose intensity of busulfan-containing preparative regimens in trials of submyeloablative transplantation.