We study the optimal allocation of machines and pallets in a class of manufacturing systems. The FMS is modeled as a closed queueing network with balanced loading of the stations. An Algorithm is developed, which exploits the properties of the throughput function and solves the allocation problem for increasing concave profit and convex cost. We also study the more general case of allocating machines and pallets among a set of FMSs. A dynamic programming approach is developed, which solves the problem with O(M$^{3}$N$^{2}$) operations.