We have previously reported a 62% response rate and 54% 1-year survival rate for patients with advanced non-small cell lung cancer (NSCLC) treated with paclitaxel (Taxol; Bristol-Myers Squibb Company, Princeton, NJ) by 24-hour infusion in combination with carboplatin, using area under the concentration-time curve dosing (FCCC 93-024). Myelosuppression proved dose limiting, but was substantially reduced by the routine use of granulocyte colony-stimulating factor during the second and subsequent cycles. Antitumor activity has been reported with minimal myelosuppression, with paclitaxel 135 and 200 mg/m2 given every 3 weeks by 1-hour infusion to patients with NSCLC. In November 1994, we initiated a phase II trial of paclitaxel 175 mg/m2 given over 1 hour, with carboplatin dosed to a fixed, targeted area under the concentration-time curve of 7.5 every 3 weeks. In the absence of grade 4 myelosuppression, paclitaxel was escalated on an intrapatient basis by 35 mg/m2 per cycle to a maximum dose of 280 mg/m2 by cycle 4. Granulocyte colony-stimulating factor was not routinely used. Eligibility stipulated advanced, measurable, chemotherapy-naive NSCLC. Of 47 patients accrued, 39 (83%) had Eastern Cooperative Oncology Group performance status 1. The median age was 64 years; 40 patients were evaluable for toxicity. Of the first 20 evaluable patients accrued (cohort A), myelosuppression was tolerable. Cumulative peripheral sensory neuropathy grade > or = 1 in 15 (75%) patients and grade 3 in six (30%), however, generally occurring at paclitaxel doses greater than 215 mg/m2, obligated removal from study of at least three patients, despite the absence of disease progression, and proved to be dose-limiting. Consequently, the protocol was revised: the starting dose of paclitaxel was reduced to 135 mg/m2, with intrapatient dose escalation of 40 mg/m2 per cycle to a maximum dose of 215 mg/m2, thus recapitulating the original dosing schema used in FCCC 93-024. To date, 25 patients have been enrolled in this second cohort (cohort B) and treatment has been better tolerated. Of 21 evaluable patients, 13 (62%) have experienced peripheral sensory neuropathy, but only one (5%) has been grade 3. Myelosuppression also has been less pronounced, with 33% grade 4 granulocytopenia and 13% grade > or = 3 thrombocytopenia in cohort B compared with 70% and 50%, respectively, in cohort A. Of the first 22 patients accrued to cohort A, 12 (55%) had major objective responses. Median event-free survival is 23 weeks and median survival is 47 weeks. Of 15 evaluable patients in cohort B, five (33%) have had major objective responses. It is too early to report survival data. In conclusion, paclitaxel by 1-hour infusion in combination with carboplatin at a fixed targeted area under the concentration-time curve of 7.5 is an active regimen in advanced NSCLC. Neurotoxicity, rather than myelosuppression, is dose and protocol limiting at paclitaxel doses exceeding 215 mg/m2.