We propose a GA-based estimation method for human muscle characteristics. We show the performance of the proposed method for fly-wheel rotating motion with elbow bending. Four main muscle parameters of Hill's two-component model are coded in a genetic code, the dynamic analytical system is made, and the GA system for evaluating muscle parameters is developed. The dynamic curves obtained by the computer simulation were found to be satisfactorily close to the experimental curves, which shows the reliability of the proposed computational model. Additionally, the dynamic analysis of constant-speed elbow flexion was simulated using the obtained parameter values. An actual experiment using an isokinetic device was performed, and both the simulation and real data were found to be in agreement with each other.