The objectives of this study were to define the pharmacodynamics of etoposide and to develop potentially useful models (1) to estimate the plasma clearance using a limited number of samples and (2) to describe the relationship between clearance and the dose-limiting toxicity. A total of 17 patients with extensive-stage small-cell lung cancer were treated with 150 mg/m2 etoposide daily for 3 consecutive days and with 100 mg/m2 cisplatin on day 3 only. Both drugs were given intravenously over 1 h. Treatment was repeated every 21 days for up to six courses. All patients were newly diagnosed (no previous chemotherapy or irradiation) and had a performance status of 0-2. Six patients achieved a complete response as confirmed by repeat bronchoscopy and five patients showed a partial response, for an overall objective response rate of 65% (95% confidence interval, 38%-87%). The median survival was 8 months (range, 1-24+ months). The dose-limiting toxicity was neutropenia. Etoposide pharmacokinetics were measured during the first course and determinations were repeated during courses 3 or 4 and 6. Complete blood counts were obtained weekly. Correlations for etoposide clearance and hematologic toxicities were evaluated for 17 initial courses and for an overall number of 33 courses. Pharmacodynamic correlations were significant for graded hematologic toxicities, as well as nadirs of leukocytes, neutrophils, and platelets for the initial courses and for all courses. To reduce the requirement for numerous blood samples, a limited sampling model was developed to estimate the area under the concentration versus time curve (AUC) with the following equation: AUC = 15.45 + 3.86 x C2 + 7.10 x C4, where C2 and C4 represent the etoposide concentrations at 2 and 4 h, respectively. The total plasma clearance was calculated as the dose divided by the AUC; correlations with toxicity were better for clearance expressed in milliliters per minute than for that expressed in milliliters per minute per square meter of body surface area. The absolute neutrophil count at the nadir (ANCn) can be estimated by the following pharmacodynamic model, which is based on 33 courses: ANCn = -0.399 + 0.024 x Ecl, where Ecl represents the etoposide clearance expressed in milliliters per minute. Further studies are necessary to validate both models prospectively.