OBJECTIVE This phase I study was undertaken to evaluate the safety and tolerability of prolonged infusional etoposide, and to evaluate its pharmacokinetic/pharmacodynamic profile in patients with advanced cancer. METHODS A group of 17 patients received a 7-day infusion of etoposide (schedule A) every 21 days at doses from 30 to 75 mg/m2 per day, and a second group of 37 patients a 21-day infusion (schedule B) every 28 days at doses from 18 to 40 mg/m2 per day. Patients had a median Karnofsky performance status (PS) of 80%, and 34 patients had no prior chemotherapy. Etoposide concentrations at steady state (Css) and other pharmacokinetic parameters (plasma clearance, CLp; area under the curve, AUC) were determined during the first treatment cycle. Correlation coefficients were calculated to measure the relationship between variables. RESULTS Myelosuppression was the major toxicity, and was associated with three deaths. The maximum tolerated dose due to neutropenia was 75 mg/m2 per day for schedule A and 40 mg/m2 per day for schedule B. There was significant interpatient pharmacokinetic variability in both infusional schedules. Even though etoposide dose levels did not significantly correlate with plasma levels, the Css was > or = 1 microgram/ml in the majority of the patients. A significant correlation between AUC and neutrophil absolute decrease was noted only in schedule B (r = 0.56, P = 0.003). There were several marginal relationships in schedule B: PS versus Css (r = 0.31, P = 0.058), PS versus AUC (r = -0.38; P = 0.058) and age versus CLp (r = -0.31, P = 0.057). CONCLUSIONS Overall, significant correlations were found for several hematologic variables and etoposide dose levels, but not with the Css values. One major problem with the application of pharmacodynamic models to predict hematologic toxicity in clinical practice is the presence of significant interpatient variability.