A unified method for computing the dynamic load-carrying capacity (DLCC) and the inverse dynamics of multiple cooperating robotic manipulators is developed in this paper. In this method, the kinematic constraints and the governing dynamic equations of the multi-robot system are formulated in the joint space based on a coordinate partitioning technique and the generalized D'Alembert's principle. This approach not only gives the minimum number of equations but also allows separate computations of the dynamics of the robots and that of the payload. The upper limit of the DLCC at any points on a given trajectory is obtained by solving a small-size linear programming problem. An iterative algorithm for evaluating the maximum allowable DLCC of the trajectory is also developed. After the DLCC is determined, the optimum distribution of actuator torques are computed by using a quadratic programming technique. The proposed method is conceptually straightforward, and it is applicable to wide varieties of multi-robot systems as well as fully parallel robots with redundant actuators.>