The diagnostic criteria for schizophrenia in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) are based on the premise that it is a discrete illness entity, in particular, distinct from the affective psychoses. This assumption has persisted for more than a century, even though patients with a diagnosis of schizophrenia show a wide diversity of symptoms and outcomes, and no biological or psychological feature has been found to be pathognomonic of the disorder. However, there has been sustained, and indeed growing, criticism of the concept. For example, writing about the diagnosis of schizophrenia more than a decade ago,2 one of Britain's most sophisticated nosological experts, Ian Brockington, enjoined "It is important to loosen the grip which the concept of 'schizophrenia' has on the minds of psychiatrists. Schizophrenia is an idea whose very essence is equivocal, a nosological category without natural boundaries, a barren hypothesis. Such a blurred concept is 'not a valid object of scientific enquiry'."3 Should Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition(DSM-V), persist with the neo-Kraepelinian concept of schizophrenia with all its defects, or should it deconstruct psychosis into its component dimensions? In this article, we will address the question by considering 2 main themes, firstly, the role of culture and ethnicity in the diagnosis of psychosis, and secondly, a life course approach to understanding psychosis. We will then discuss whether more progress would be achieved in DSM-V by abandoning the familiar categorical system and instead moving to a dimensional system which rates both developmental impairment and symptom factor scores. However, we will begin by briefly reviewing the recent history of the classification of the psychoses.