The pelvis is the predominant site of failure following radical radiotherapy (RT) for locally advanced cervical cancer. We report the results of phase I-II studies on 200 patients with bulky (greater than or equal to 5 cm) carcinoma of the cervix. Patients were treated between 1981 and 1988 on sequential protocols of concurrent chemoradiation to establish an acceptable treatment regimen. RT with daily or partially hyperfractionated pelvic (n = 154) or pelvic plus paraaortic (n = 46) fields was given by continuous (n = 154) or split course (n = 46) regimens. Infusional fluorouracil (5-FU) in a dose of 1 g/m2/day was given on the first and last 4 days of a 5-week course of continuous RT, or with both halves of split course RT. Seventy-eight patients received bolus mitomycin C (Mit-C), 6 mg/m2, once or twice with the start of the 5-FU infusion. The median external RT dose was 46 Gy (range 40 to 65 Gy) followed in 90% (n = 181) by a single intracavitary application of 40 Gy using a linear source of cesium-137. Median follow up is 2.5 years (range 0.6 to 6.9 years) and is sufficient to reliably estimate late toxicities. Acute toxicities were transient oral mucositis (13), RT interruption for enteritis (7), febrile neutropenia (3), and thrombocytopenic tumor bleed (1). Serious late toxicities resulted in death in 3 patients and occurred in bladder in 6 and in bowel in 25, including 8 patients with tumor recurrence. The incidence of late bowel toxicity correlated with the specific therapy given and decreased with each successive protocol. On logistic regression the only treatment variable showing a statistically significant effect on complications was the use of Mit-C (P = 0.0053). Pelvic RT and 5-FU alone produced fewer complications, only 4/105, than historically seen with standard pelvic RT alone. Three-year pelvic control and survival rates were 85 and 71% respectively in stage Ib/II (n = 100) and 50 and 42% in stage III/IV (n = 100). Encouraged by these results and decreased toxicity, we have begun a phase III study to determine whether the addition of concurrent 5-FU to continuous partially hyperfractionated pelvic RT improves local control and survival.