Background: Cisplatin-based regimens are standard first-line chemotherapy for ED-SCLC. In patients unfit for cisplatin due to advanced age or poor performance status (PS), carboplatin plus etoposide (CE) is as effective as cisplatin plus etoposide (JCOG9702 trial). Carboplatin plus irinotecan (CI) and carboplatin plus amrubicin (CA) are promising new carboplatin-based regimens identified in our previous studies (NJLCG0405 and NJCLG0701). Accordingly, we conducted this randomized phase 2 study to identify the appropriate regimen for comparison with CE in future phase 3 trials. Methods: Chemotherapy-naïve ED-SCLC patients were randomly assigned to receive 4–6 cycles of carboplatin (area under the curve [AUC] 5.0, day 1) plus irinotecan (70 mg/m², days 1 and 8) every 3 weeks (CI arm) or carboplatin (AUC 4.0, day 1) plus amrubicin (35 mg/m², days 1–3) every 3 weeks (CA arm). The primary endpoint was the overall response rate (ORR). Secondary endpoints were progression-free survival (PFS), overall survival(OS), and toxicity. Assuming that an ORR of 80% in eligible patients indicates potential usefulness and an ORR of 60% is the lower limit of interest, the target sample size was 35 patients in each arm (alpha, 0.05; beta, 0.20). Results: Between December 2009 and March 2013, 71 patients were enrolled. One patient in each arm did not receive any protocol treatment due to rapid disease progression. Characteristics of treated patients were as follows: median age, 70 years (range 51–84 years); proportion of males, 84%. The ORRs were 79% (95% confidence interval [CI]: 62–91) and 89% (95%CI: 73–97), median PFS were 5.1 and 6.3 months (hazard ratio [HR] = 0.51, 95%CI: 0.30–0.85, p = 0.01), and median OS were 14.9 and 15.9 months in the CI and CA arms, respectively. Toxicities of grade 3 or higher severity were neutropenia (CI, 53% and CA, 89%), anemia (CI, 26% and CA, 20%), thrombocytopenia (CI, 18% and CA, 14%), and febrile neutropenia (CI, 12% and CA, 29%). No treatment-related deaths were observed. Conclusions: CA was numerically effective than CI in chemo-naïve ED-SCLC patients, with acceptable toxicity. CA could be selected for future phase 3 trials. Clinical trial information: UMIN000008970.