For two component xerographic development systems the toner concentration and the average toner residence time in the developer sump are desired to be within optimal ranges for acceptable image quality. The toner concentration and the average toner residence time are coupled such that independent ranges are not typically achievable, and so a tradeoff must be managed. Also the required toner for any given job is highly variable and not known well in advance, yet significantly impacts the two aforementioned parameters. In this paper the authors propose and demonstrate a model predictive control design framework to optimally manage toner throughput to achieve performance objectives.