OBJECTIVE: To evaluate the effect of extensive lymphadenectomy on survival in patients with gastric cancer. DESIGN: Retrospective analysis SETTING: University Hospital, Norway. SUBJECTS: 183 patients with stomach cancer resected for cure during the time period 1980-90. INTERVENTIONS: 78 patients had an R1- and 105 patients and R2 resection. 124 patients were treated by total gastrectomy, 5 by proximal--and 54 by distal resection. MAIN OUTCOME MEASURES: Morbidity, mortality and long term survival. RESULTS: The morbidity was 33% (60/183), of which 39 (21%) were general complications (pneumonia, thrombosis, or cardiovascular disease). 14 patients died postoperatively (8%). By logistic regression analysis we found that splenectomy was the only variable associated with both morbidity and immediate postoperative mortality. Five year survival was 39% for patients who had undergone curative resections, 30% for patients who had had an R1 resection, and 47% for those who had had an R2 resection. By multivariate analysis (Cox) we found that N-classification (TNM), tumour diameter of less than 45 mm, type of lymph node dissection (R2) and operation period (after 1984) correlated with improved survival. CONCLUSION: Extensive lymph node dissection improves survival without increasing morbidity or postoperative mortality.