The pharmacodynamics of etoposide phosphate (Etopophos; Bristol-Myers Squibb Company, Princeton, NJ), a water-soluble prodrug of etoposide, was evaluated in 39 patients with solid tumors after a 30-minute intravenous infusion of escalating doses (equivalent to 50 to 175 mg/m2 of etoposide) on a day 1, 3, and 5 schedule of treatment. Serial blood samples were collected at predose and throughout the 32 hours following day 1 of treatment to determine the area under the plasma concentration-time curve (AUC) of etoposide phosphate and etoposide. Hematology profiles and serum chemistries were determined at predose and twice weekly for approximately 3 weeks after each treatment cycle. Both linear and nonlinear pharmacodynamic models were used to evaluate the relationship between hematologic toxicity and etoposide AUC and patient factors (age, gender, performance status, prior radiation therapy, prior chemotherapy, baseline albumin, bilirubin, alkaline phosphatase, creatinine, leukocyte count, granulocyte count). Etoposide phosphate was converted rapidly to etoposide in vivo. The ratio of the etoposide phosphate AUC to that of etoposide was < or = 1.2% indicating that etoposide was the main species in the systemic circulation. Myelosuppression was the dose-limiting toxicity, and significant decreases in white blood cell and granulocyte counts were noted. Hematologic toxicity was best described by a stepwise linear regression model consisting of etoposide AUC, serum albumin, and bilirubin. In summary, hematologic toxicity produced by the intravenous administration of etoposide phosphate correlates significantly with etoposide AUC and patient factors (baseline serum albumin and bilirubin) in cancer patients.