The extended lymphadenectomy (D2) was recently introduced in several guidelines as the optimal treatment for gastric cancer, based only on the 15-year follow-up results of the Dutch randomised trial, while the British Medical Research Council (MRC) study failed to demonstrate a survival benefit over the more limited D1 dissection. The Italian Gastric Cancer Study Group randomised controlled trial (RCT) was also undertaken to compare D1 versus D2 gastrectomy, and a tendency to improve survival in patients with advanced resectable disease (pT > 1N+) was documented despite negative results in the entire patient population. Now we present the 15-year follow-up results of survival and gastric cancer-related mortality. Between June 1998 and December 2006, eligible patients with gastric cancer who signed the informed consent were randomised at 5 centres to either D1 or D2 gastrectomy. Intraoperative randomisation was implemented centrally by phone call. Primary outcome was overall survival (OS); secondary end-points were disease-specific survival, postoperative morbidity and mortality. Analyses were by intention to treat. Strict quality control measures for surgery, lymph node removal, pathology and patient follow-up were implemented and monitored. Registration number: ISRCTN11154654 (http://www.controlled-trials.com). A total of 267 eligible patients were assigned to either D1 (133 patients) or D2 (134) procedure. Median follow-up time was 16.76 years. Analyses were done both in overall patient population and in pT > 1N+. One hundred patients (38.5) were alive without recurrence. OS and disease-specific survival (DSS) were very high in both arms. In overall population, they were not different between D1 and D2 arm (51.3% vs. 46.8% and 65% vs. 67% respectively, p = 0.31 and p = 0.94). DSS was significantly higher after D2 in pT > 1N+ patients (29.4% vs. 51.4%, p = 0.035). OS and DSS were better after D1 in patients older than 70 years (p = 0.003 and p = 0.006). DSS was higher after D1 also in early stages (p = 0.01). After 15-year follow up, despite no relevant difference in overall population, DSS and gastric cancer-related mortality of patients with advanced disease and lymph node metastases are improved by D2 procedure. Further data available from this trial suggest that D1 procedure should be preferably used in older patients and in early disease. As accurate detection of advanced diseases can be currently provided by adequate preoperative workup in referral centres, D2 procedure should be recommended in these cases. Piedmont Regional fund for Finalized Healthy Research Project, Application 2003 for data collection. •In overall population, the survival benefit of D2 dissection is not evident.•In resectable locally advanced GC, D2 dissection improves disease-specific survival.•Preoperative staging could identify locally advanced disease in referral centres.•D2 dissection is not necessary in early gastric cancer.•In older patients, D2 procedure is not superior and can be dangerous.