The use of biomarkers is increasing both in acute and chronic disease epidemiology, but the rationale for their introduction is not always firmly established (e.g., when and how their use is scientifically justifiable and cost effective). The use of biomarkers should be considered within the context of causal models in epidemiology, and of the intertwining of causation and pathogenesis. Unlike infectious diseases, for cancer and cardiovascular disease external "necessary" causes have not been identified. Thus, the classification of cancer and other chronic diseases cannot be based on unequivocal criteria such as the "etiologic" classification of infectious diseases. As far as morphology is concerned, "neoplasia" and "anaplasia" are attributes of cancer that cannot be defined in a straightforward way. Tissue pathologies are minimal and difficult to differentiate from normal tissue in some cancers but are obvious in others. From a mechanistic point of view, unless molecular biology discovers specific mechanistic steps in carcinogenesis, which indicate the existence of "necessary" events in carcinogenesis, we cannot adopt an unequivocal definition of cancer. The potential contribution of biomarkers to the elucidation of the pathogenetic process should be considered in the light of such uncertainties. There is a range of indications for biomarkers, from the use of very specific measurements aimed at single molecules, to measurements indicating cumulative exposure to agents with the same mechanism of action. The potential uses of markers in chronic disease epidemiology include (1) exposure assessment in cases in which traditional epidemiologic tools are insufficient (particularly for low doses and low risks); (2) multiple exposures or mixtures, in which the aim is to disentangle the etiologic role of single agents; (3) estimation of the total burden of exposure to chemicals having the same mechanistic target; (4) investigation of pathogenetic mechanisms, and (5) study of individual susceptibility (e.g., metabolic polymorphism, DNA repair).