One hundred fifty-nine patients with unresectable but localized pancreatic cancer, as defined at exploratory laparotomy, were treated at the Mayo Clinic between February 1974 to April 1985. Postoperative therapy consisted of 4000 to 6000 cGy external beam irradiation (XRT) alone in 122 patients or 4500 to 5500 cGy XRT in combination with an intraoperative electron boost in 37. In addition, 132 (both groups) received 5-fluorouracil (5-FU) chemotherapy. Local control (LC) at 1 year was 82% with XRT + intraoperative radiation therapy (IORT) versus 48% with XRT and 66% versus 20% at 2 years respectively (P less than 0.0005). Due to the high incidence of hematogenous and/or peritoneal spread in both groups (abdominal failure in 54 and 56% of patients at risk), the decreased frequency of local progression did not translate into an improved survival. Neither median nor long-term survival of the two treatment groups (XRT versus XRT + IORT) was statistically different (median 12.6 months versus 13.4 months, P = 0.25). With tumor arising in the head of the pancreas, survival at 2 years was 18% as opposed to 0% for other locations (P less than 0.01). On the basis of a Cox multivariate analysis, no other treatment or prognostic factor significantly altered survival. Until the problem with systemic failure (usually abdominal) can be resolved, the median and long-term survival of patients with pancreatic carcinoma is likely to remain unchanged. Since IORT appears to improve local control, we will continue to utilize IORT in phase 1, 2 studies which also attempt to decrease the incidence of abdominal failures. Even with IORT + XRT combinations, the incidence of local progression is excessive and radiation dose modifiers need to be evaluated.