The traditional approach to conducting Phase I studies of anticancer drugs is to select a starting dosage for humans based on preclinical data (e.g., mg equivalent of 1/10 LD10 in mice), then empirically escalate dosages in cohorts of patients until the maximum tolerated dosage (MTD) is established. More recently, NCl and EORTC investigators have advocated the use of pharmacokinetic data from preclinical studies to facilitate more rapid dose escalation (e.g., double the dose until the area under the concentration-time curve [AUC] in humans equals the AUC in mice at the LD10). The present paper describes a strategy which builds on the above approach, by extending the application of pharmacokinetic principles to systematically escalate systemic exposure (AUC) instead of dosage in Phase I trials. Human trials are initiated at whatever patient-specific dosage is required to achieve an AUC equal to 1/10 the AUC in mice at the LD10, such that three patients at the first treatment level might receive three different dosages. If no dose-limiting toxicity is observed, the next cohort of patients receives whatever dosage is required to achieve 2 x AUC of the first dosage level, with AUC escalation continuing until the maximum tolerate systemic exposure (MTSE) is reached. By escalating systemic exposure instead of dosage, one adjusts for interpatient pharmacokinetic variability. This strategy will permit more rapid and precise dosage escalations and, more importantly, it should more precisely establish the maximum level of treatment intensity for future Phase II trials.