A parallel processing algorithm based on the recursive dynamics formulation presented in a companion paper  is developed for multiprocessor implementation. Lagrange multipliers associated with cut-joint constraints for closed loop systems are eliminated systematically, resulting to a minimal set of equations of motion. Concurrent generation of the system inertia matrix and the generalized force vector is exploited. A new computational structure for dynamic analysis is proposed for real-time parallel processing. Real-time simulation of a vehicle is performed to illustrate efficiency and effectiveness of the algorithm, even for interactive man-in-the-loop simulation.