Despite the progress in molecular genetics, the genes responsible for the development of bipolar disorder (BPD) and schizophrenia have not yet been identified. This failure can be attributed to an ambiguous phenotypic description and several variations in the genetic transmission of these diseases. There is a growing consensus that an endophenotype approach may be utilized to overcome the difficulties regarding the phenotypic description and facilitate the identification of the susceptibility or protective genes. The endophenotypes which can be defined as subclinical vulnerability markers may assist in the identification of the genetic underpinnings of psychiatric disorders regardless of the disease status. This approach may provide well-defined phenotypes having a stronger relationship with the pathophysiology and genetic etiology than with the diagnostic categories themselves. An endophenotype may be an inherited neurophysiological, neuropsychological, cognitive, neuroanatomical, biochemical or endocrinological trait. Nevertheless, it must be 1) associated with illness, 2) present in nonaffected family members at a higher rate than in the general population, 3) present within the normal population to a lesser extent and 4) state-independent. Besides increasing the power of genetic analysis in BPD and schizophrenia, an endophenotype approach may help in reshaping the classical nosological systems and diagnostic categories. Lastly, it may have additional use in psychiatry, including the development of suitable animal models for these disorders. In this article, the rationale for the use of endophenotypes in genetic studies of BPD and schizophrenia is discussed and the proposed candidate endophenotypes for both disorders are reviewed.