Background: Neuroendocrine carcinoma (NEC) arising in the digestive system is a rare and high-grade malignant tumor. Treatment guidelines for advanced NEC recommend platinum-based chemotherapy regimens. Etoposide plus cisplatin (EP) and irinotecan plus cisplatin (IP) are commonly used as community standard regimens for NEC. However, it is unknown which is more effective. This phase III study aimed to identify more effective regimen between EP and IP in terms of overall survival (OS) for patients with advanced NEC. Methods: Eligibility criteria included chemotherapy-naïve patients with recurrent or unresectable, histologically confirmed NEC (WHO2010) arising from the gastrointestinal tract (GI), hepatobiliary and pancreas (HBP), ECOG PS of 0-1, age of 20–75 years. In the EP arm, etoposide (100 mg/m² on days 1, 2 and 3) and cisplatin (80 mg/m² on day 1) were administered every 3 weeks. In the IP arm, irinotecan (60 mg/m² on days 1, 8 and 15) and cisplatin (60 mg/m² on day 1) were repeated every 4 weeks. The primary endpoint was OS. Total of 140 patients (114 events) were required to detect a hazard ratio (HR) of 0.67 (median OS; 8/12 months in inferior/superior arms) with a two-sided alpha of 10% and power of 70%. Due to a rapid accrual, the sample size was increased to 170 (power of 80%). Pathology was centrally reviewed after the start of treatment. Results: Of all enrolled 170 patients (GI: 100, HBP: 70) from Aug 2014 to March 2020, median OS was 12.5 months in EP and 10.9 months in IP (HR 1.043 [a ratio of EP to IP], 90% CI, 0.794-1.370, p= 0.797). Median progression-free survival (PFS) was 5.6 months in EP and 5.1 months in IP (HR 1.060, 95% CI, 0.777-1.445). Response rate (RR) was 54.5% (42/77) in EP and 52.5% (42/80) in IP among patients with target lesions. No interactions with OS, PFS, or RR were seen in pre-planned subgroup analyses including primary organs or pathological findings. The common Grade 3-4 AEs (EP/IP) was neutropenia (91.5%/53.7%), leukocytopenia (61.0%/30.5%), and febrile neutropenia (26.8%/12.2%). Conclusions: A superiority was not demonstrated, which indicates that both EP and IP remain standard first-line chemotherapy. Clinical trial information: UMIN000014795.