Summary Background Phase I/II clinical trials of S-1 plus cisplatin for advanced gastric cancer have yielded good responses and the treatment was well tolerated. In this S-1 Plus cisplatin versus S-1 In RCT In the Treatment for Stomach cancer (SPIRITS) trial, we aimed to verify that overall survival was better in patients with advanced gastric cancer treated with S-1 plus cisplatin than with S-1 alone. Methods In this phase III trial, chemotherapy-naive patients with advanced gastric cancer were enrolled betweeen March 26, 2002, and Nov 30, 2004, at 38 centres in Japan, and randomly assigned to S-1 plus cisplatin or S-1 alone. In patients assigned to S-1 plus cisplatin, S-1 (40–60 mg depending on patient's body surface area) was given orally, twice daily for 3 consecutive weeks, and 60 mg/m2 cisplatin was given intravenously on day 8, followed by a 2-week rest period, within a 5-week cycle. Those assigned to S-1 alone received the same dose of S-1 twice daily for 4 consecutive weeks, followed by a 2-week rest period, within a 6-week cycle. The primary endpoint was overall survival. Secondary endpoints were progression-free survival, proportions of responders, and safety. Analysis was by intention to treat. This trial is registered with ClinicalTrials.gov , number NCT00150670. Findings 305 patients were enrolled; seven patients were ineligible or withdrew consent, therefore, 148 patients were assigned to S-1 plus cisplatin and 150 patients were assigned to S-1 alone. Median overall survival was significantly longer in patients assigned to S-1 plus cisplatin (13·0 months IQR 7·6–21·9) than in those assigned to S-1 alone (11·0 months 5·6–19·8; hazard ratio for death, 0·77; 95% CI 0·61–0·98; p=0·04). Progression-free survival was significantly longer in patients assigned to S-1 plus cisplatin than in those assigned to S-1 alone (median progression-free survival 6·0 months 3·3–12·9 vs 4·0 months 2·1–6·8; p<0·0001). Additionally, of 87 patients assigned S-1 plus cisplatin who had target tumours, one patient had a complete response and 46 patients had partial responses, ie, a total of 54% (range 43–65). Of 106 patients assigned S-1 alone who had target tumours, one patient had a complete response and 32 had partial responses, ie, a total of 31% (23–41). We recorded more grade 3 or 4 adverse events including leucopenia, neutropenia, anaemia, nausea, and anorexia, in the group assigned to S-1 plus cisplatin than in the group assigned to S-1 alone. There were no treatment-related deaths in either group. Interpretation S-1 plus cisplatin holds promise of becoming a standard first-line treatment for patients with advanced gastric cancer.