Schizophrenia is one of the most severe and disabling psychiatric disorders. Antipsychotic drugs offer considerable benefits in controlling symptoms and preventing relapse. The strategy for the present review of clinical trials was to ask 'What are the features of schizophrenia and the existing treatments of the illness that have implications for future clinical trials'? Six key facts were identified.First, schizophrenia is genetically 'complex'. Trials may benefit from designs including genetically related illnesses, by focussing on cross-cutting aspects of the phenotype such as psychosis or cognitive dysfunction, and by collecting information on possible moderators and mediators of treatment response.Second, schizophrenia affects multiple neurotransmitter systems. Multiple signalling pathways may need to be considered, with different time courses of response. Outcome measures from clinical trials could be collected at more frequent intervals, particularly in the early phase of response.Third, the clinical features used to define the illness are a mix of symptoms and social-occupational dysfunction, yet treatment response is often defined only by changes in symptoms. Multiple measures of functioning need to be collected at baseline and at the endpoint of trials. Consensus definitions for response, remission, relapse, recovery and recurrence need to be developed.Fourth, schizophrenia is often highly disabling. Linking treatment response in clinical trials to measures of quality-adjusted life-years will allow comparison with other medical illnesses using common metrics.Fifth, the general health and care of individuals with schizophrenia is often poor. 'Complex' interventions, which include, but are not limited to, antipsychotic medications, need to be designed and tested for the problems facing these patients.Finally, large gaps exist between clinical trials, practice guidelines and patterns of practice. Trials need to be designed to investigate widely used approaches such as antipsychotic polypharmacy, where actual practice diverges from evidence-based guidelines.