A review was carried out of the published literature on survival rates for cancer in relation to patterns of organisation of medical care, specifically treatment at specialist centres or at hospitals treating larger numbers of patients and treatment by protocol, usually within the context of a clinical trial. Centralised referral or entry to trials was frequently associated with a higher survival rate, particularly for the less common cancers, and was never found to be associated with a lower survival rate. Few studies were identified for any one cancer site and some antedated current methods of treatment. At a time when the health service in the United Kingdom is undergoing far-reaching organisational change, further research is needed to establish the most beneficial patterns of care for people with cancer. Population-based cancer registries are an invaluable source of data for such studies.