A new incomplete repair model is introduced that differs from previous models of this type by not assuming that repair is complete during long intervals, e.g. “overnight” intervals of 12–24 h. The model was used to assess the risk of myelopathy resulting from continuous hyperfractionated accelerated radiotherapy treatment (CHART) in light of recent experimental data on the rat spinal cord [2]. Model calculations employing biexponential repair kinetics showed that CHART treatments might result in a higher myelopathy risk than an equal dose given in conventional 2-Gy fractions if the parameters obtained from the animal data hold. The probability of observing what has been reported for CHART [5] was determined in computer simulations for different variance scenarios. The chance to observe four myelopathies in the 74 cervical cord patients was estimated to range between 25 and 62%, while the probability to see 0 in 68 thoracic cord patients ranged from 48 to 27%. These numbers were derived from reasonable assumptions about the repair kinetics (e.g. 60% of damage repaired with a half-time of 8 h) so that the over-all probability to observe 4 74 and 0 68 was maximized, and depending on the scenario fell in the range 12–17%. Finally, from these simulations a myelopathy risk of approximately 0.3–1.2% is predicted for the currently employed maximal CHART dose to the spinal cord, i.e. 42 Gy. We conclude that the CHART experience is not compatible with the new experimental data ( p<5%). Incomplete repair is unlikely to be the sole reason for the unexpected toxicity of CHART ( p⩽17%). However, the compounding effect of incomplete repair could be minimized by rearranging the modified CHART schedule (with unchanged dose, dose per fraction and overall treatment time) which would result in an increase of the tolerance dose by approximately 10%.