Simulated small pedigrees (2 parents, 4 offspring) were used to illustrate the applications and limitations of a "model choice" approach designed to detect genetic heterogeneity in familial diseases. While it is possible to identify groups of pedigrees which have different genetic causes for quanitative phenotypic trait(s), theoretical limitations on discriminating between 4 single locus models exist for certain pedigree structures. These limitations originate from the overlapping phenotypic predictions of the various genetic models. Such limitations must be carefully considered in the design of genetic studies. Studies aimed at detecting genetic heterogeneity in familial diseases should limit the different genetic models being considered and tailor the sampling strategy to avoid collecting pedigrees which are non-informative for certain comparisons.