Clearly, there is a need for further elucidation of the role in carcinogenesis of dietary factors. On the epidemiologic level, studies of the last few years indicate the need to investigate further the potential role of fats as related to all the major sites of cancer, to distinguish the effects of various fractions of fiber as well as of fibers deriving from fruits, vegetables and grains, and to examine more closely carotenoids as opposed to retinoids to explore further the proposition that although carotenoids may reduce risk for a number of cancers, retinoids may play no part. Future studies must also reflect more carefully the human experience regarding diet. Because humans do not ingest foods singly, but rather as groups, patterns of intake of groups of foods and nutrients may provide insight into types of diet that affect risk. In nutritional science, there is an increasing appreciation of the complexity of interaction of nutrients both in absorption and in metabolism; future epidemiologic studies of biologic materials as well as of nutrient interactions may be beneficial. Interindividual differences in response to diet which may identify subgroups who are more or less susceptible to particular dietary factors should also be examined. Since dietary exposures affect virtually everyone, the public health implications of even small differences in risk are important. For example, a well substantiated relative risk of 1.2 for a nutrient associated with breast cancer is not trivial. Speculation as to the proportion of total cancer attributable to diet is so tenuous as to be almost frivolous.(ABSTRACT TRUNCATED AT 250 WORDS)