Of the 2790 patients with gastric carcinoma who underwent surgery between January 1978 and December 1993, there were 224 who underwent a D2 resection for cancer of the proximal third of the stomach. Survival in these patients was retrospectively analyzed with special reference to the survival benefit of pancreaticosplenectomy (PS). The D2 resection involved a splenectomy in 129 patients, a PS in 38 patients, and neither in 57 patients, with 5-year survivals of 74%, 59%, and 91%, respectively. Although these differences in survival were attributable to the apparent deviation in the incidence of prognostic variables, no significant survival benefit of PS was observed for patients with the same clinical stage disease (stages II and III) or the same modal status (n1 and n2). Because PS was associated with a significant increase in the incidence of anastomotic dehiscence and length of hospital stay, the dismal survival benefit implies that it should not be performed routinely with the D2 resection but should be employed only for lesions with direct invasion of the pancreas.