We have designed a combined treatment strategy of bifractionated split course radiotherapy (RT) and concomitant chemotherapy (CT) to try to improve the results of RT in inoperable cervical carcinoma. After evaluation, patients were submitted to further radical surgery or additional RT-CT depending on the treatment results. Between January 1988 and January 1992, 25 patients with non metastatic inoperable disease entered in the protocol. The stage of the disease was: T3N0 4 patients, T3 with hydronephrosis seven patients, T3N1 12 patients, and T4N0 two patients. Nineteen patients received two courses of CT with fluorouracil (F), cisplatin (P) with or without etoposide. Pelvic RT was given twice daily (two fractions of 3 Gy) on days 1, 3, 15 and 17. A combination of F 400 mg/m2/d and P 15 mg/m2/d in continuous infusion with oral etoposide (100 mg/d) and hydroxyurea (500 mg/d) in 11 patients was delivered concomitantly on days 1-3 and 14-17. A clinical and radiological evaluation was performed four weeks later. Patients with objective response underwent radical hysterectomy (group A) and those with incomplete response received additional RT-CT protocol (group B). All patients had endocavitary brachytherapy at the end of treatment. After two cycles of CT there were four PR in 19 patients and 5 failures. After RT-CT there were 12 CR (48%) and eight PR. There was a relationship between disease status after RT-CT and response to initial CT in those 19 patients who received the neoadjuvant CT. Fifteen patients were in group A, six of whom had no histologically active disease in the post-operative sample. However all 15 patients were rendered free of disease. Ten patients were in group B, five of whom attained the clinical CR status. In total, 20 of 25 patients (80%) were in CR at the end of treatment. Six patients experienced pelvic recurrence and two patients distant metastases. Four of the five patients with incomplete response had evolutive disease. The overall survival is 60% and 36% at 1 and 2 years respectively after a median follow-up of 22 months (14-48 m). The protocol was tolerable. These results compare favorably with those of conventional RT and warrant further evaluation.