Thirty-six patients underwent curative resection of a primary pancreatic carcinoma from January 1977 to September 1987; 26 had Whipple resections, seven had total pancreatectomies, and three had distal pancreatectomies. Twenty-six patients manifested recurrent disease, four died of intercurrent disease, and six were apparently cured. Median survival was 11.5 months with actuarial survival at 2 and 5 years of 32% and 17%, respectively. Of the eventual recurrences, 19% were local only (pancreatic bed, regional nodes, adjacent organs, and immediately adjacent peritoneum) and 73% had a component of local failure. All patients failing did so with a component in the intraabdominal cavity. Peritoneal (42%) and hepatic failures (62%) were common. Extraabdominal metastases were documented in only 27%, but never as a sole site. Fourteen patient and tumor characteristics were evaluated for any relationships with failure or survival. No single variable independently predicted for local failure. However, a group of three (age > 60 years, T2 or T3 stage, and location of tumor in the body or tail) was associated with a substantial local failure risk (85% of all patients with local failure). Multivariate analysis showed that low tumor grade (P = O.oOZ), female sex (P = 0.002), and adjuvant radiation (P = 0.02) were all independent predictors of prolonged survival. Ten patients were treated in an adjacent setting. Those given 55 Gy or greater had improved local control (50% versus 25%) and cure (33% versus none) when compared with patients treated to lower doses. The authors conclude that local failure after curative resection remains a significant problem and further efforts to improve local control are warranted. However, peritoneal and hepatic relapses occur frequently. Thus, adjuvant treatment strategies using wide-field radiation techniques or intraperitoneal therapy, in combination with local tumor bed irradiation and chemotherapy, should be explored. Cancer 66:56-61,1990. ANCREATIC CARCINOMA remains a devastating P problem for the patient as well as those charged with its management. By recent estimates, this malignancy will account for 3% of the total cancer incidence in adults and 5% of the total deaths, taking 24,500 lives in the US in 1988.' Moreover, the incidence has risen over the last 40 years.2 Even in the highly selected subgroup that undergoes "curative" resection, the expected 5-year survival is in the range of 0% to 18%,3 with median survivals of 10 to 14 month^.^,^ For the remainder of patients, approximately one third will have unresectable locoregional disease only and the rest will have metastases at pre~entation.~ In patients who have undergone complete resection, a previous randomized study from the Gastrointestinal Tumor Study Group (GITSG) as well as a follow-up non-randomized registration study have shown a survival advantage for patients treated with postoperative combined radiation and 5-fluorouracil (5-FU).637 In light of this 56