We have attempted to clarify the nature of heterogeneity associated with affective disorder. Three categories of such heterogeneity must be recognized, ie, heterogeneity in symptomatology, heterogeneity in treatment response, and heterogeneity in etiology. The three are not necessarily related, and evidence has been presented that it is unlikely that relationships between them are simple. It follows from the evidence presented that inference about heterogeneity of etiology cannot be drawn from either heterogeneity of symptomatology or treatment response. Some of the possible relationships between these three constructs are discussed, and problems faced by the investigator in dealing with such complex relationships are pointed out. Consideration of the complex relationships between etiologies, symptoms, and treatment makes it desirable to approach research on depressive illness from a multivariable, multidimensional viewpoint.