As a result of strong selection in closed gene pools, dog breeds represent populations of highly related animals. Prominent founder-effects are responsible for inherited diseases occurring in particular breeds, and each breed may have several breed-related diseases, often with a high incidence. Such inherited diseases are a major problem in purebred dog populations, and probably threaten their survival. On the basis of pedigree information held by the National Kennel Clubs and reliable medical data of a representative longitudinal cohort, estimates have been made of the relative risks of dichotomous disease traits in all combined breeding stock. This approach is independent of assumptipns about modes of inheritance or thresholds. In a cohort study, all the common ancestors of the cases are selected and their degree of relatedness to both cases and controls is estimated. The ancestors which are positively associated with the dichotomous disease trait are selected on the basis of scores of relatedness. To reduce the number of parameters, while maintaining maximal informativeness, a principal component analysis is applied. Finally a logistic model, based on the principal components and the case control definitions, describes the most likely pattern of the passage of genetic risk factors down the generations. Estimates of relatedness to seven highly related ancestors were sufficient to describe the distribution of disease in a population of Dutch labrador retrievers. This approach may be used for genetic counselling for any clinical phenotypically dichotomous trait in such a highly related population of companion animals, and may also help to identify suitable dogs for molecular studies of the underlying defect.