This paper proposes a method for using dynamic programming to solve the optimal chiller sequencing problem and to eliminate the deficiencies of conventional methods. The coefficient of performance of the chiller is adopted as the objective function because it is concave. The Lagrangian method determines the optimal chiller loading in each feasible state. The potential performance of the proposed method is examined with reference to three example systems. The proposed method consumes much less power than the conventional method, and is very appropriate for application in air-conditioning systems.