This report investigates relationships between the pharmacokinetics and pharmacodynamics of high-dose alkylators used for the treatment of primary breast cancer. Eighty-five women with primary breast cancer involving >or=10 lymph nodes received four cycles of standard-dose chemotherapy followed by a high-dose regimen consisting of: cyclophosphamide (1875 mg/m(2) once daily x 3), cisplatin (165 mg/m(2) given over 72 h), carmustine (600 mg/m(2)), and stem cell transplantation. Dosages were attenuated in patients whose body weight exceeded their calculated ideal weight by >20%. Pharmacokinetics of the high-dose chemotherapeutic agents were evaluated in each patient by collection and analysis of serial blood samples. Area under the concentration time curve (AUC) for cyclophosphamide and carmustine was highly variable (>10-fold inter-patient range) with coefficients of variation > 50%, in contrast to cisplatin exposures (2-fold range; coefficient of variation 12%). The dosing method for overweight patients resulted in significantly lower systemic exposure to cisplatin (P = 0.035). The parent cyclophosphamide clearance on the 1st day of administration was significantly higher in patients who experienced acute cardiac toxicity (n = 5; P = 0.011), whereas carmustine disposition was not found to be different in those developing pulmonary toxicity (n = 25; P = 0.96). Kaplan-Meier analysis (median follow-up of 5.9 years) demonstrated that patients with lower cyclophosphamide AUC (faster parent drug clearance to potentially cytotoxic compounds) survived longer (P = 0.031). Inter-individual differences in the pharmacokinetic disposition of high-dose chemotherapy may explain variability in both response and toxicity. Prospective strategies, which attempt to individualize AUC, should be evaluated in this setting.