Methods for the design and analysis of allelic association studies in the presence of two comorbid disorders are discussed. These methods are applicable to population-based (i.e., case-control) designs. We first develop probability models that represent pathways to the comorbidity of two disorders when it is hypothesized that at least one of the two disorders is associated with a specific allele. These potential pathways are illustrated with the specific example of the association of the human leptin (OB) gene with obesity and major depressive disorders in humans. We then discuss methods of design and analysis using the well-known methods of log-linear analysis [Bishop et al., 1975: "Discrete Multivariate Analysis." M.I.T. Press, Cambridge, MA] to differentiate between these pathways. With the increasing focus in psychiatric genetics on both association studies and pathways to comorbidity we anticipate that these methods will have wide applications.