This study used the frequency of food-use data collected previously in the Western New York Diet Study, a case-control study of diet and cancer at various sites, to develop measures of dietary patterns for subsequent use in epidemiologic research. Factor analysis was applied to frequency-of-use data of 110 foods for 1475 male and 780 female control subjects. Nine dietary-pattern factors were extracted for each gender. Correlational analyses examined relationships with intakes of nutrients of concern to cancer epidemiologists--energy, total fat, dietary fiber, and vitamins A and C. Findings indicate that the usual measure for assessing dietary risks, quantile of nutrient intake, reflects several dietary patterns differing in their composition of cancer risk factors. Furthermore, patterns may relate to such variables as age, ethnicity, and income, which are themselves related to cancer. This approach may provide a comprehensive measure of dietary exposure for epidemiologic research.