PURPOSE: To determine, in a randomized comparison, whether the addition of paclitaxel to etoposide and cisplatin improves the time to progression and overall survival in patients with extensive small-cell lung cancer (SCLC) compared with standard etoposide and cisplatin and to compare the regimens' toxicity. PATIENTS AND METHODS: Eligible patients (N=587) with untreated extensive SCLC were randomly assigned to receive either cisplatin 80 mg/m2 on day 1 and etoposide 80 mg/m2 on days 1 through 3 administered every 3 weeks for six cycles (EP) or cisplatin 80 mg/m2 on day 1, paclitaxel 175 mg/m2 over 4 hours on day 1, and etoposide 80 mg/m2 on days 1 to 3 followed by recombinant human granulocyte colony-stimulating factor on days 4 to 18 administered every 3 weeks for six cycles (PET). RESULTS: Reporting of demographics, response, and survival included 565 patients, of whom 282 were randomly assigned to receive EP and 283 were assigned to receive PET. Overall response rates were 68% for the EP arm and 75% for the PET arm. Median failure-free survival time was 5.9 months for the EP arm and 6 months for the PET arm (P = .179). Median overall survival time was 9.9 months for patients on EP and 10.6 months for patients on PET (P = .169). Toxic deaths occurred in 2.4% of the patients on EP and 6.5% of patients on PET. CONCLUSION: PET did not improve the time to progression or survival in patients with extensive SCLC compared with EP alone and was associated with unacceptable toxicity.