OBJECTIVES: F100 is a fortified milk-based food used as a standard therapy for severe malnutrition. Our aim was to assess whether optimal combinations of local foods in Ghana, Bangladesh, and Latin America could achieve the nutrient density levels of F100 and, if that was unachievable, to identify the key limiting nutrients. MATERIALS AND METHODS: In all of the analyses, a published list of population-specific foods (n = 59) and maximum food portions were used. The nutrient densities of individual foods were calculated to identify nutrient-dense foods and potential limiting nutrients. The feasibility of formulating diets that achieved the nutrient density of F100, for each region individually and for all regions combined, was explored by the use of linear programming analysis (4 models), which minimized the total grams of food while respecting model constraints on diet energy and nutrients, phytate:zinc molar ratio, and maximum food portion sizes. RESULTS: For vitamin E, riboflavin, zinc, and copper, fewer than 5 foods achieved the nutrient density of F100. All 4 linear programming model solutions were unfeasible, indicating that in all regions, community-based home-prepared rehabilitation diets for severely malnourished children will not achieve the nutrient densities of F100. In model solutions, the densities of vitamin E, riboflavin, zinc, copper, calcium, thiamin, and niacin were unachievable, indicating they are potential key problem nutrients. CONCLUSIONS: Optimal combinations of local foods are unlikely to achieve the nutrient density of F100, especially for vitamin E, riboflavin, thiamin, niacin, zinc, calcium, and copper. Before home-prepared rehabilitation diets with nutrient densities lower than those of F100 are promoted, it is important to establish their clinical efficacy.