The impact of lymphadenectomy in prognosis and staging in periampullary malignancies remains largely undefined. We examined all pancreaticoduodenectomies for periampullary carcinomas in the SEER cancer registry from 1993 through 2003. Overall, 5465 pancreaticoduodenectomies for nonmetastatic periampullary carcinomas were identified. The cohort was comprised of 62.5% pancreatic, 18.9% ampullary, 11.6% distal bile duct, and 7.0% duodenal cancers. A linear association between the number of lymph nodes (LNs) examined and overall survival was observed overall and for pancreas and ampullary cancers for node-negative (N0) disease. Median survival for all patients with localized, N0 disease improved from 24 to 31 months, with sampling of a minimum of 10 LNs, whereas 2 and 5-year survival improved from 52 and 29%, with <10 nodes examined to 58 and 37% with 10+ nodes examined (P<0.001). A 1-month median survival advantage was seen in patients with node-positive disease when more than 10 lymph nodes examined (15 versus 16 months, P<0.001). Significantly better median survival and cure rates are observed after pancreaticoduodenectomy for localized periampullary adenocarcinoma when a minimum of 10 lymph nodes are examined. This benefit likely represents more accurate staging. To optimize the prognostic accuracy and prevent stage migration errors in multicenter trials a minimum of 10 lymph nodes should be obtained and examined before the determination of node-negative disease.