The limited success of genetic studies of major depression has raised questions concerning the definition of genetically relevant phenotypes. This paper presents strategies to improve the phenotypic definition of major depression by proposing endophenotypes at two levels: First, dissecting the depressive phenotype into key components results in narrow definitions of putative psychopathological endophenotypes: mood bias toward negative emotions, impaired reward function, impaired learning and memory, neurovegetative signs, impaired diurnal variation, impaired executive cognitive function, psychomotor change, and increased stress sensitivity. A review of the recent literature on neurobiological and genetic findings associated with these components is given. Second, the most consistent heritable biological markers of major depression are proposed as biological endophenotypes for genetic studies: REM sleep abnormalities, functional and structural brain abnormalities, dysfunctions in serotonergic, catecholaminergic, hypothalamic-pituitary-adrenocortical axis, and CRH systems, and intracellular signal transduction endophenotypes. The associations among the psychopathological and biological endophenotypes are discussed with respect to specificity, temporal stability, heritability, familiality, and clinical and biological plausibility. Finally, the case is made for the development of a new classification system in order to reduce the heterogeneity of depression representing a major impediment to elucidating the genetic and neurobiological basis of this common, severe, and often life-threatening illness.