Dynamic; programming techniques appropriate to optimizing a maintenance function are evaluated to demonstrate the feasibility of applying them to large-scale systems involving many equipments. The objective is to determine the proper number, mix, and level of military and/or contractor maintenance personnel, and to apportion expenditures on other maintenance resources, such as test equipment, computer programs, and manuals. The tasks to which the model was applied (drawn from comprehensive data on a developmental AF environment) considered maintenance of 15 digital display consoles, and used personnel, computer programs, computer manuals, procedural manuals (including circuit diagrams), and test equipment as the variable resources. The results yield the mix of resources that achieves a given mean-time-to-repair at minimum cost. Alternatively, the results can yield the trade-off of cost and mean-time-to-repair. Constraints on particular variables can be injected into the procedure—and, in fact, have the effect of reducing required computer runs—but must be carefully restricted to those that realistically reflect the actual situation. The model not only realistically represents the maintenance function, but also can be readily expanded to other equipments by adding stages and data to the tables.