We develop here a model for restriction fragment length distributions based on DNA dimer frequencies in humans. Mean fragment lengths are computed for known restriction enzymes. This model is tested using data from the hybridization of a series of arbitrary single-copy DNA probes screened with a set of restriction enzymes. The fit to the model appears good. We apply the model to the problem of how much DNA is scanned by a set of enzymes. This result is then further applied to optimizing the search for insertion/deletion DNA-polymorphisms.