The kinetics of repair of sublethal lesions in the spinal cord was assessed in detail using a rodent model. Experiments were designed to obtain the fractionation sensitivity, alpha/beta, and to quantify the kinetics of repair after a clinically relevant fraction size. Pairs of 2-Gy fractions at intervals ranging from 0 to 24 h were given from Monday through Friday, to cumulative doses of 40-84 Gy. In addition, two groups of animals received 1.5 or 1.2 Gy twice a day at 8-h intervals, 5 days a week, to total doses of 66-90 Gy and 67.2-98.4 Gy, respectively. All irradiations were followed by a top-up dose of 16 Gy. Analysis of the experimental data revealed that a bi-exponential repair model fit the experimental data significantly better than did the mono-exponential model (p = 0.002). The repair half times obtained were 0.7 (0.2-1.3) h and 3.8 (2.6-4.9) h, respectively. The proportion of injury repaired by the longer half time was estimated to be 0.62 (0.37-0.86). The data showed that delivering 2 fractions per day at 6- or 8-h intervals instead of one per day led to a 16.5% (11.8-21.1%) and 13.5% (9.1-17.8%) reduction in the tolerance, respectively. Finally, the results indicated that when incomplete repair between fractions was accounted for, the linear-quadratic (LQ) model was valid in describing fractionation response down to 1.2 Gy per fraction.