There is a need for a simple, rapid assay for predicting normal tissue reactions in radiotherapy patients to reduce morbidity in sensitive patients and to allow dose escalation in resistant cases. Towards this goal we have investigated the gamma-ray sensitivity of lymphocytes from 16 breast cancer patients who had shown an exaggerated acute or late radiation reaction ('overreaction') of normal tissues after radiotherapy, using chromosome damage (dicentrics) as the endpoint because of its close relationship with cell killing. The use of a low dose-rate (LDR; 0.31 cGy min-1) was found to be better than a high dose-rate (170 cGy min-1) in discriminating between over-reactors and controls, as predicted (and here confirmed) from previous studies on ataxia-telangiectasia (A-T) homozygotes and heterozygotes. Five of seven patients with excessive early skin reactions (e.g. erythema, moist desquamation) showed abnormal radiosensitivity, manifested either as aberration yields above the control range after LDR exposure or as less sparing than controls. The average LDR yield for early over-reactions was significantly higher than for controls (p = 0.009) and average sparing was less (p = 0.0002). Two of 10 patients with late complications (fibrosis, telangiectasia) had LDR yields above the control range, but the average yield for late over-reactors was not significantly above that of controls. Unexpectedly, two patients (one early, one late reaction) had LDR aberration yields below the control range. Quantitatively our results are consistent with the notion that over-reacting breast cancer patients are carriers of the A-T gene. Pilot studies on controls showed that the sparing effect of LDR irradiation was increased by lowering the dose-rate to 0.13 cGy min-1 and by using micronuclei rather than metaphase damage as the endpoint. These modifications to the protocol will be used in a large-scale prospective study.