The Radiation Therapy Oncology Group (RTOG) initiated a phase I/II study of intraoperative radiotherapy (IORT) in advanced or recurrent rectal cancer to assess therapeutic efficacy, toxicity, and establish quality control guidelines prior to beginning a phase III trial. From October 1985 through December 1989, 87 patients with histologically proven adenocarcinoma of the rectum or rectosigmoid with recurrent/persistent disease after surgery or those primarily inoperable were entered by 14 institutions. Of 86 evaluable patients, 42 patients received IORT either alone (n = 15) or in combination with external beam (n = 27). Local control was dependent on the amount of residual disease prior to IORT, with 2-year actuarial local control of 77% if no gross residual disease remained vs. 10% with gross residual disease (P = 0.001). For the recurrent/residual group (n = 33), this observation was also significant with a 2-year actuarial local control rate of 64% if no gross residual remained vs. 10% with gross residual disease (P = 0.004). Local control translated into an improved survival for all patients and the recurrent/residual group with 2-year actuarial survival of 88% and 89% if no gross residual disease remained vs. 48% and 45% with gross residual disease, respectively (P = .0005, 0.006). Six patients (14.6%) experienced four grade 3 and three grade 4 complications as a possible result of IORT during follow-up with a 2-year actuarial risk of major complications of 16%. We conclude that IORT is feasible within a cooperative group and can be performed with acceptable complication rates. A phase III trial to demonstrate a therapeutic advantage for IORT over external beam alone is currently in progress.