This article describes a mathematical model that takes an existing food intake pattern and develops a new pattern that meets a specified set of nutrition recommendations with the minimum change possible. The model examines multiple recommendations simultaneously and considers foods as they are currently eaten, so it can provide practitioners with new insights about strategies for implementing recommendations. The model shows serving units per day by food group in both existing and new intake patterns and the recommendations responsible for changes. Recommendations of the National Research Council are used, and sodium-restricted (< or = 2,400 mg) and sodium-unrestricted patterns are compared. Food intake data are from 915 nonpregnant, nonlactating women 19 to 50 years old who participated in the US Department of Agriculture's 1985 Continuing Survey of Food Intake by Individuals. The sodium-restricted pattern contains proportionately more dairy products, fruit, and red meat and proportionately less grain products, fish, and vegetables than the unrestricted-sodium pattern. Nutrition recommendations responsible for changes in the existing pattern include those for saturated fatty acids (< 10% kcal), carbohydrate (> or = 55% kcal), zinc (100% of Recommended Dietary Allowance [RDA]), potassium (> or = 3,500 mg/day), and vitamin E (100% RDA), in addition to sodium. Implications of the sodium recommendation for food selection and preparation, especially of food mixtures and commercially prepared products, are discussed.