Abstract A retrospective review of palliative outcome of gastrojejunostomy in patients with pancreatic cancer was conducted. Eighty-one patients were analyzed in two groups depending on duodenal patency. Forty-five patients (group I) had no evidence of duodenal obstruction. Thirty-six patients (group II) had evidence of impingement on the duodenum by the pancreatic cancer. A third subset of patients was also studied for outcome. These 21 patients (five group I and 16 group II) had nausea and vomiting as major symptoms and were judged to have the most to gain from gastrojejunostomy. Patients were catagorized by outcome. Poor outcome was defined as either death during the hospitalization for gastrojejunostomy or death within 30 days of operation even if the patient left the hospital. Risk for poor outcome depended on group. In group I, 18 of 45 patients (40%) had a poor outcome compared with 25 of 36 (70%) patients in group II (p < 0.001). Nineteen of the 21 (90%) patients with nausea and vomiting had a poor outcome. It is an unfortunate paradox that the more patients need gastrojejunostomy for pancreatic cancer, the less likely they are to have a favorable outcome. Gastric outlet obstruction in pancreatic cancer appears to be a terminal event. A prospective study is needed to see if any true palliation of vomiting can be affected in these patients.