AbstractAll knowledge is necessarily provisional and grows by trial and by error elimination (Popper). Scientific theories are provisional and are not dogmatic; they provide explanations and predictions which are potentially falsifiable (by observation or experiment). Many other meaningful theories (linguistic, historical, political, sociological, psychoanalytic) are not scientific, as they do not give rise to predictions which are potentially falsifiable.Scientific explanations are preferable for two reasons. Firstly, because they are testable (and not because they are necessarily correct); secondly, because they are relatively value free: given time, most rational men will come to similar conclusions on the same scientific evidence; if there is a conflict, this will usually lead to the collection of new data to decide which is the preferred explanation.Scientific method should be used to help identify and solve community health problems. The results of such scientific studies will usually suggest a need for social action and the precise action to be taken will be influenced by political, economic and social values as well as by the scientific evidence.Scientific method proceeds by defining a problem situation, and by posing one or more specific questions which, if answered, would help to resolve the problem situation. In some instances, questions can be posed and answered by observation and critical analysis of the problem situation; in the health context, a problem situation will often arise out of a conflict of opinion as to the comparative effectiveness of two “treatments” or health plans. To solve problems such as this, it will be necessary to carry out randomised controlled trials. The elements to be randomised can be patients, hospitals, wards, areas or districts, trainee health workers; the “treatments” could be medical, surgical, community health strategies health education policies, etc.If there is no scientific evidence of the comparative effectiveness of different aspects of medicine, then health care planning will become more and more dominated by the political pressures to reduce health costs. On the other hand, if we can provide scientific evidence of comparative effectiveness, then for any planning (or treatment) decision we should be able to compare the medical consequences with the economic consequences and then (on socio‐political grounds), strike a rational balance bet