"Two assumptions are commonly made in the estimation of genetic risk: (1) that the seven specific loci in the mouse constitute a suitable basis for extrapolation to genetic disease in humans, and (2) that mutations are induced by radiation damage (energy-loss events leading to double-stranded damage) occurring within the gene and are induced linearly with dose, at least at low doses. Recent evidence on the mutability of repeat sequences is reviewed that suggests that neither of these assumptions is as well founded as we like to think."