From August 1985 through September 1989, 284 patients with advanced pelvic malignancies were entered into a trial (RTOG 8502) of palliative split course radiation (4440 cGy in three courses of 1480 cGy/2 days/4 fractions with a rest of 2-4 weeks between courses). The initial 148 patients were part of a Phase II acceptable response rate and minimal acute or late toxicity (IJRBP 17:659-662, 1989). The present analysis is a report of the subsequent 136 patients randomized between rest intervals of 2 weeks versus 4 weeks to determine if length of rest would influence tumor response or patient toxicity. The patients were stratified for performance status (Karnofsky Performance Status) and histology. The patients were evenly matched for age and sex. There was a trend toward increased acute toxicity incidence in patients with shorter rest interval (5/68 versus 0/68; p = .07). Late toxicity was not significantly different between the two groups. Decreasing the interval between courses did not result in a significant improvement in tumor response (CR+PR = 34% vs. 26%, p = n.s.). More patients in the 2 week groups completed all three courses (72% vs. 63%). Not surprisingly, patients completing cell three courses had a significantly higher overall response rate than for patients completing less than three courses (42% vs. 5%) and higher complete response rate (17% vs. 1%). A multivariate analysis indicated performance status as the significant predictor for number of courses completed. For Karnofsky Performance Status greater than or equal to 80, the survival at 12 months was 40% for the 2 week interval and 25% for the 4 week interval. Performance status and histology were the only significant variables in a multivariate analysis of survival.