A series of 191 patients submitted to adjuvant radiotherapy after surgery for rectal and rectosigmoid carcinoma from January, 1975, to December, 1990, has been analyzed to evaluate the incidence of high grade small bowel late toxicity (grades III/IV according to RTOG/EORTC scoring system). Surgical approach was abdominoperineal Miles resection in 92 patients, while a sphincter preserving bowel resection was performed in 99. The total radiation dose to the pelvis ranged from 40 to 60 Gy in 4 to 8 weeks, with conventional fractionation. Three different beams arrangements were used: two sagittal parallel opposite AP-PA fields in 106 patients, three fields (one posterior sagittal plus two parallel lateral fields) in 56, four fields (box technique) in 29. Fourteen patients (7.3%) developed sequelae grades III/IV: three of them died of toxicity. Average free interval between radiation and complication was 23 months (range: 4-87). The following risk factors were investigated: sex, age, type of surgery on primary rectosigmoid cancer, previous abdominal or pelvic surgery, radiation technique, treated volume, administered radiation dose. Dose was calculated as BED (time corrected biologically equivalent dose) according to the linear quadratic model. The only factors significantly related to late intestinal complications were the beam arrangement and, consequently, the treated volume. Detailed analysis showed that radiation sequelae developed in 12/106 (11.3%) patients treated with the two sagittal fields technique, while small bowel toxicity was observed in only 2/85 (2.3%) patients treated with the three--or four--fields technique. The difference is stastically significant (p < 0.05). Another significant correlation was the ratio between treated volume and late complications incidence observed.(ABSTRACT TRUNCATED AT 250 WORDS)