Abstract
In this paper, concepts of dynamic programming are used within a discrete time Markovian model for the development of a graded population. Optimal recruitment and transition patterns are determined by minimizing expected discrepancies between actual states and preferred goals.