This paper describes a modeling and nonlinear control design method for a manipulator based on multibody dynamics (MBD). Each arm of the robot has seven degrees of freedom, similar to a human arm. The arm is a redundant manipulator, so it is generally difficult to control. To derive an accurate model of the robot arm, it is straightforward to set up three-dimensional equations of motion using MBD. We propose two controllers using the MBD model. One is an LQI controller with inverse dynamics. The other one is a nonlinear controller using the structure of the MBD model. We tested the effectiveness of our proposed method by a simulation and experiment. As the results, the tracking performance of the nonlinear controller based on the MBD model was about 18% better than the controller of the LQI controller with inverse dynamics.