Linking specific chromosomal regions with the abnormal genes responsible for major psychiatric disorders is rendered particularly difficult because of three factors: diagnostic uncertainty; the possibility that some psychiatric disorders are caused by the concerted action of more than one gene; and the possibility that similar disorders may result from alternative genetic abnormalities. The approach proposed here is to concentrate initially on the highly focused problem of resolving the question of genetic heterogeneity of a given disorder. To this end, it is recommended (a) that studies concentrate on pedigrees large enough to contain at least two nuclear families with several affected individuals, and (b) that, within them, work be confined to an intensive examination of such small nuclear families containing an affected parent and at least two affected offspring with a fairly unmistakable phenotype-bipolar disorder, type I. By using molecular and genetic techniques designed to extract maximal information from this relatively small number of affected individuals, it should be possible to determine whether or not bipolar disorder is genetically heterogeneous. This is a limited but very important question, and the technology is presently available to answer it. The answer should, in turn, guide subsequent approaches to identification of the relevant gene or genes.