In metastatic or recurrent cervical cancer, cisplatin-based chemotherapy is standard. The JCOG0505 randomized phase III trial evaluated the clinical benefits of carboplatin-based regimen. Eligible patients had metastatic or recurrent cervical cancer and had ≤ one platinum-containing treatment and no prior taxane. Patients were randomly assigned either to conventional paclitaxel plus cisplatin (TP; paclitaxel 135 mg/m(2) over 24 hours on day 1 and cisplatin 50 mg/m(2) on day 2, repeated every 3 weeks) or paclitaxel plus carboplatin (TC; paclitaxel 175 mg/m(2) over 3 hours and carboplatin area under curve 5 mg/mL/min on day 1, repeated every 3 weeks). Primary end point was overall survival (OS). Planned sample size was 250 patients to confirm the noninferiority of TC versus TP with the threshold hazard ratio (HR) of 1.29. Between February 2006 and November 2009, 253 patients were enrolled. The HR of OS was 0.994 (90% CI, 0.79 to 1.25; noninferiority P = .032 by stratified Cox regression). Median OS was 18.3 months with TP versus 17.5 months with TC. Among patients who had not received prior cisplatin, OS was shorter with TC (13.0 v 23.2 months; HR, 1.571; 95% CI, 1.06 to 2.32). One treatment-related death occurred with TC. Proportion of nonhospitalization periods was significantly longer with TC (P < .001). TC was noninferior to TP and should be a standard treatment option for metastatic or recurrent cervical cancer. However, cisplatin is still the key drug for patients who have not received platinum agents.