This prospective randomized trial was undertaken to determine the added efficacy of (32)P in treating locally advanced unresectable pancreatic cancer. Thirty patients with biopsy proven locally advanced unresectable adenocarcinoma of the pancreas were assessable after receiving 5-fluorouracil and radiation therapy with or without (32)P, followed by gemcitabine. Intratumoral (32)P dose was determined by tumor size and volume and was administered at months 0, 1, 2, 6, 7, and 8. Tumor cross-sectional area and liquefaction were determined at intervals by computed tomography scan. Tumor liquefaction occurred in 78% of patients receiving (32)P and in 8% of patients not receiving (32)P, although tumor cross-sectional area did not decrease. Serious adverse events occurred more often per patient for patients receiving (32)P (4.2 +/- 3.1 vs. 1.8 +/- 1.9; p = 0.03) leading to more hospitalizations. Death was because of disease progression (23 patients), gastrointenstinal hemorrhage (4 patients), and stroke (1 patient). One patient not receiving (32)P and one receiving (32)P are alive at 28 and 13 months, respectively. (32)P did not prolong survival (7.4 +/- 5.5 months with (32)P vs. 11.5 +/- 8.0 months without (32)P, p = 0.16). (32)P promoted tumor liquefaction, but did not decrease tumor size. Intratumoral (32)P was associated with more serious adverse events and did not improve survival for locally advanced unresectable pancreatic cancer.