The principal purpose of this study was to evaluate relationships between paclitaxel plasma steady-state concentration (Css) and both disease outcome and toxicity in patients with non-small cell lung cancer (NSCLC) treated with paclitaxel and cisplatin in an Eastern Cooperative Oncology Group (ECOG) Phase III study E5592. Chemotherapy-naive patients with stage IIIb and IV NSCLC were randomized to treatment with either 75 mg/m2 cisplatin i.v. on day 1 and 100 mg/m2 etoposide i.v. on days 1-3 (EC arm) or 75 mg/m2 cisplatin i.v. combined with either a low dose of paclitaxel (135 mg/m2, 24-h i.v. infusion; PC arm) or a higher dose of paclitaxel (250 mg/m2 i.v., 24-h i.v. infusion) with granulocyte colony-stimulating factor (PCG arm). End-of-24-h-infusion paclitaxel concentrations, which have been demonstrated to be nearly equal to CssS on this schedule, were obtained during the first and second courses in patients on the PC and PCG arms. Relationships between the average paclitaxel Css (Css,avg) and the best response to treatment, time to treatment failure (TTF), survival, and worst grade of leukopenia and neurotoxicity were evaluated by univariate analysis. A multivariate model was used to assess the influence of paclitaxel Css in conjunction with other potentially relevant patient variables that may affect disease outcome, including the paclitaxel treatment arm, age, sex, performance status, weight loss during the previous 6 months, and disease stage. Paclitaxel Css in both courses 1 and 2 were obtained in 71 patients treated with PC and 75 patients treated with PCG. Although Css,avgS in patients treated with PC and PCG were significantly different (median, 0.32 versus 0.81 micromol/liter; P < 0.0001), response rates were not (33.8 versus 26.7%; P = 0.3719). In addition, there were no differences between the PC and PCG arms in TTF (median, 5.1 versus 5.5 months, P = 0.6201) or survival (median, 11.6 versus 11.3 months, P = 0.7173). Combined analysis of paclitaxel concentrations from both treatment arms revealed no significant difference in paclitaxel Css,avg between responders and nonresponders [median, 0.40 (range, 0.16-1.6) micromol/liter versus 0.55 (range, 0.11-3.6)], and Css,avgS were similar in patients segregated according to whether they had a complete response, partial response, stable disease, or progressive disease as their best response to treatment (P = 0.7612). In addition, the relationship between Css,avg and TTF was weak (r2 = 0.00003, P = 0.94), as was the relationship between Css,avg and survival (P = 0.1267). With regard to the principal toxicities, neither the propensity to develop neuromuscular and neurosensory toxicity nor the worst grade of these adverse effects were related to Css,avg (P = 0.5000 and 0.2033, respectively); however, the relationship between Css,avg and the worst grade of leukopenia experienced was marginally significant (P = 0.0796). In a multivariate model, neither the combined effect of relevant demographic and stratification variables nor paclitaxel Css,avg predicted for either response (P = 0.1544) or TTF (P = 0.2574), whereas the combined effect of all covariates predicted for survival (P = 0.0249). With regard to individual covariates, a lower disease stage (stage IIIb) was the only significant positive determinant of response (P = 0.0173), female sex was the only significant favorable predictor for TTF (P = 0.0195), and a lower ECOG performance status (= 0) was the only significant positive determinant of survival (P = 0.0121) in the multivariate model. In summary, paclitaxel Css,avg was not a determinant of response, TTF, or survival in patients with advanced NSCLC treated with paclitaxel as a 24-h i.v. infusion combined with cisplatin. On the basis of both the clinical and pharmacodynamic results of E5592, there is no compelling reason to treat patients with advanced NSCLC with paclitaxel on a 24-h i.v. schedule at doses of > 135 mg/m2 in combination with cisplatin, although highe