Bringing epidemiology and toxicology together to better understand cause and effect relationships requires attention to several interconnected problems: problems of commitment, complexity, and of communication. The most fundamental of these is commitment as it is reflected in the basic purpose of environmental epidemiology. The purpose of epidemiology is not to prove cause/effect relationships, and not only because scientific proof is elusive. The purpose of epidemiology is to acquire knowledge about the determinants and distributions of disease and to apply that knowledge to improve public health. A key problem, therefore, is how much and what kinds of evidence are sufficient to warrant public health (typically preventive) actions? The assessment of available evidence lays the foundation for the problem of complexity: relevant evidence arrives from toxicologic and epidemiological investigations, and reflects the acquisition of knowledge from many levels of scientific understanding: molecular, cellular, tissue, organ systems, complete organisms (man and mouse), relationships between individuals, and on to social and political processes that may impact human health. How to combine evidence from several levels of understanding will require the effective communication of current methodological practices. The practice of causal inference in contemporary environmental epidemiology, for example, relies upon three largely qualitative methods: systematic narrative re views, criteria-based inference methods, and (increasingly) meta-analysis. These methods are described as they are currently used in practice and several key problems in that practice are highlighted including the relevance to public health practice of toxicological evidence.