Stochastic kernels are used in this paper to investigate intra-distribution dynamics in the world per capita intakes of proteins and fats. The analysis of actual transitions over the last 40 years indicates that lagging countries improved their position relative to the leading. Long-run (steady-state) distributions have been obtained using estimated intake change models. These distributions have been compared to “virtual” ones revealing that the income elasticity of demand or equivalently the rate of growth in per capita income does have a strong influence on the dispersion of intakes at the steady-state.