The current system of classification of mental health diagnoses has been widely criticised. In particular, the diagnosis of schizophrenia has been highlighted as being essentially meaningless. Issues surrounding its validity and utility have been raised. In response to this, an alternative to the current system has been proposed: the psychosis subgroups. These subgroups, informed by the biopsychosocial, stress vulnerability model and Cognitive Behavioural Therapy approaches, are drug-related psychosis, traumatic psychosis, stress sensitivity psychosis and anxiety psychosis. This thesis explores validity and utility of classifications within mental health, through the investigation of the psychosis subgroups. Three studies were designed to explore the validity and utility of the psychosis subgroups from the perspective of three main stakeholders in the use of mental health classification: researchers, service users and clinicians. Study 1 explored the validity of the psychosis subgroups using a quantitative cross sectional study. A clinical sample was used to investigate concurrent validity. The hypothesis was not upheld and possible reasons for this are discussed. Study 2 explored Professional Experts and Experts by Experience opinions about the validity and utility of psychosis subgroups using a Delphi study. A lack of consensus in many areas was found. Further descriptive data identified divergent opinions held by participants. Study 3 used Grounded Theory to explore the opinions of Experts by Experience. A model of "uncertain knowing" in relation to the causes and explanations of psychosis was proposed. The implications and practical uses of this model are discussed. This thesis provides a pragmatic, comprehensive and systematic exploration of the validity and utility of the psychosis subgroups. Whilst no conclusive evidence for the psychosis subgroups was found, the Grounded Theory model and methodology used may have practical implications for further investigation and exploration of mental health classification.