After decades of neuroscientific research and taxonomic endeavour in psychiatry the quest for biologic markers specific enough to accommodate nosologic categories has not succeeded. Yet, neurophysiology and neuroimaging have developed pwoerful tools to investigate brain function. An immense amount of data has been accumulated regarding normal and pathologic information processing, cognition, emotion and other domains. Some have been correlated with genes underpinning diseases and are candidate endophenotypes. These stand at an intermediate level between genes and phenotype. They encompass several kinds of dysfunctions or abnormalities in brain structure. Rather than matching to singular diagnostic categories, as we devise them today, the same endophenotype is usually shared by distinct pathologic entities. Assuming tha they reflect specif dysfunctions this raises critical questions regarding the DSM way of classifying mental disorders and to the understanding of the neurobiologic phenomena underlying them. It is the purpose of this presentation to discuss these questions and review some of the data, including our own, concerning event-related potentials endophenotypes of psychosis with special focus on the schizophrenia-bipolar dichotomy and present.