BACKGROUND: The value of second-line therapy for metastatic gastric cancer is unclear. So far there are no randomised phase III data comparing second-line chemotherapy to best supportive care (BSC). In this prospective, multicenter, open label, randomised phase III study we compared irinotecan to BSC to evaluate the impact on survival of second-line chemotherapy. METHODS: Eligible patients (pts) had metastatic or locally advanced gastro-oesophageal junction or gastric adenocarcinoma, objective tumour progression during or within 6months after first-line chemotherapy and ECOG performance status 0-2. Stratification for time of progression after first-line therapy, ECOG PS and pretreatment secured even distribution of important prognostic factors. TREATMENT: Arm A: Irinotecan 250mg/m(2)q3w (first cycle) to be increased to 350mg/m(2), depending on toxicity. Arm B: BSC. FINDINGS: Between 10/2002 and 12/2006 40 pts were randomised. The study was closed prematurely due to poor accrual. Responsefor arm A (19 pts evaluable): No objective responses, SD 53%, PD 47%. Improvement of tumour related symptoms: Arm A 50% of pts, arm B 7%. Overall Survival: (all events in 40 pts have occurred): The hazard ratio for death was reduced to 0.48 (95%CI 0.25-0.92) in the irinotecan-arm (p=0.012). Median survival arm A: 4.0months (95% CI 3.6-7.5), arm B: 2.4months (95% CI 1.7-4.9). INTERPRETATION: Irinotecan as second-line chemotherapy significantly prolongs overall survival compared to BSC in the studied pts. Second-line chemotherapy can now be considered as a proven treatment option for metastatic or locally advanced gastric cancer. FUNDING: The study was supported by a research grant from Aventis and Pfizer.