Ten years ago, Tsuang et al. (1 ) stated that “psychiatric genetics had reached a point where the sophistication of available experimental tools such as molecular genetics technologies and statistical procedures has surpassed the ability to describe relevant phenotypes.” Indeed, the early enthusiasm about the role that molecular genetics would play in our understanding of mental disorders was tempered in the late 1980s by the failure to reproduce the previously observed linkage between bipolar affective disorder and schizophrenia (2 ,3 ). This may be partly attributed to the obstacles that hampered efforts to identify the genes responsible for a particular complex disorder, including unknown mode of inheritance, genetic heterogeneity, phenocopies, incomplete penetrance, and variable expressivity (4 ,5 ). However, an even larger obstacle for the identification of the genes underlying genetic vulnerability of psychiatric disorders is our inability to define the heritable phenotype. Although reliable diagnostic criteria and structured psychiatric interviews have been used to identify probands, little is known about their genetic validity. Within apparently affected subjects, various types of phenotypic misclassification reduce the power of linkage studies because of phenocopies or genetic heterogeneity. Furthermore, our inability to identify non-affected subjects carrying vulnerable genes from a population of apparently unaffected subjects or controls, as a result of incomplete penetrance, also reduces the power of association studies. The failure to identify genetic susceptibility loci in psychiatric disorders may thus be partly responsible for inadequate phenotypic definition.