Questions concerning the scope, theory and method of ecological research in psychiatry are discussed in the light of some prominent findings, which adumbrate the present state of knowledge of environmental influences on mental health. The epidemiological model is taken as a frame of reference. The special problems which arise in testing hypotheses, and their solution, are illustrated by the example of the ‘ecological fallacy’. Here, an attempt is made to progress from simple correlations between ecological variables and defined illness-rates, and speculation about their significance, to a careful analysis of unavoidable conditions, and of their relative importa