This paper presents results of a ten-year study done in the field of modeling and solving the problem of human nutrition. In solving the problem of institutional menu planning, a two phase approach is proposed: Phase I — comprising of the determination of how many times the given meal has to be used in the given planning period, and Phase II — comprising the precise definition of each of the daily menus used, taking care of all nutritional, economical and other constraints pertinent to the highly complex domain of human nutrition. The Phase I model reduces to linear, integer goal programming model with about 250 unknowns and about 300 constraints. The model used in Phase II is a nonlinear, 0-1 goal programming model, consisting of, typically, from 50000 to 100000 of unknowns and of about 200 constraints. The solution method is based on modified, iterative Monte Carlo procedure using a hypergeometric distribution for the random variable representing the unknown. The model has been successfully used for a number of years for programming and planning of human nutrition in military environment. However, the usefulness of the approach appears to be of some importance in the case of the hospitals and other institutions feeding a large number of persons.