Genetic dissection of diseases is one of the epoch-making achievements in modern medicine. Positional cloning is a key method to isolate disease-related genes. For positional cloning, there are two conventional methods: family-based studies and case-control studies. In this review, I would like to describe several family-based studies on single gene diseases which I had conducted including those of Akita diabetic mice, systemic carnitine deficiency and Hartnup disease. The study of systemic carnitine deficiency underscored a potential power of the "Carrier state." Furthermore, cultural and public health practices in Japan such as preservation of umbilical cords and mother and child passbooks enabled us to conduct linkage analysis even 20 years after the deaths of affected patients in Hartnup disease. For multifactorial diseases, I present three family-based studies: intracranial aneurysm, moyamoya and arteriovenous malformation. Finally, I discuss on theoretical issues concerning the relationship among odds ratio, phenocopy rate and penetrance by formulating a single-locus dominant association model. Analysis of the model predicted a notion that a large odds ratio facilitates familial clustering of multifactorial diseases and vice versa is the case. Furthermore, the analysis predicted that genetic markers for screening should have odds ratio >/= eight to maintain similar qualities commonly required for clinical tests. Collectively, the analysis predicted a two-stage study design composed of linkage analysis based on a family study and subsequent replication by a case-control association study is more rational than the currently used two-independent case-control design. This newly proposed method is expected to provide polymorphisms, which have large odds ratios, requiring only minimum research budgets.