The analysis of nine Utah families that were ascertained for clusters of breast cancer cases is reported. Segregation analysis of an inherited susceptibility to breast cancer shows two distinct maximum likelihood solutions that have almost equal likelihood. One model indicates that most females had zero risk for breast cancer, but 10% of the female population had risks much greater than the Utah age-specific incidence rates. The other model indicates that most females have a risk defined by the Utah rates for breast cancer, but a rare dominant gene is segregating for increased susceptibility to breast cancer. Our analysis shows that linkage results under the two models are consistent in sign but not in magnitude. No evidence for linkage was found with the 14 marker loci examined. In addition to demonstrating distortion of linkage results from ignoring sporadic cases, this analysis shows the inherent difficulty of obtaining parameter estimates for segregation analysis when families are ascertained from a cluster of cases.