This study investigates the association of microsatellite polymorphisms in XRCC1, XRCC3 and XRCC5 with the development of late radiation-induced radiotherapy reactions and examines the correlation between these microsatellites and cancer incidence. Sixty-two women with cervical or endometrial cancer treated with radiotherapy were included in the study. According to the CTCAEv3.0 scale, 22 patients showed late adverse radiotherapy reactions (grade 2 or more). PCR on lymphocyte DNA followed by automated fragment analysis was performed to examine the number of tandem repeat units at each locus. No significant association was found between the repeat length at any of the microsatellites in XRCC1, XRCC3 or XRCC5 and the incidence of late radiotherapy complications. Since higher odds ratios (ORs) were found for the rare XRCC1 [AC] 11 and [AC] 21 repeats (OR ؍ 2.65, P ؍ 0.325 and OR ؍ 8.67, P ؍ 0.093, respectively), the possible involvement of these small and large repeats in clinical radiosensitivity cannot be completely ruled out. When specific numbers of repeats were examined, no significant correlation was found between the microsatellite repeat length in XRCC1 and XRCC5 and cancer incidence. A weak correlation between XRCC3 [AC] 16 homozygotes and cancer incidence was found (OR ؍ 2.56, P ؍ 0.055). A large-scale multicenter study of cancer patients with a high number of radiosensitive individuals is needed to clarify the value of rare polymorphic microsatellite repeats in XRCC1 and XRCC3 as a biomarker of clinical radiosensitivity or increased cancer risk.