This paper addresses ecological studies in public health research in terms of the logic of their analysis. It makes several distinctions between studies based on ecological and individual units. First, it identifies the variables common to both types of study and those particular to ecological studies. Second, it shows how ecological and individual units combine in two classes: unmixed (purely ecological, purely individual) and mixed. Third, it details how the relationships among and between individual and grouped units (expressed in terms of regression coefficients between independent and dependent variables) yield four coefficients: for all individual members; for all groups; for all individuals within each group; and for all individuals within groups (a weighted average). Equipped with an understanding of the dimensions involved at ecological and individual levels and of the relationships between them, researchers are in a position to exploit the public health potential of the ecological approach.