Model-based pose estimation of human motion in video is one of important tasks in computer vision. This paper proposes a novel approach using an orthogonal simulated annealing to effectively solve the pose estimation problem. The investigated problem is formulated as a parameter optimization problem and an objective function based on silhouette features is used. The high performance of orthogonal simulated annealing is compared with those of the genetic algorithm and simulated annealing. Effectiveness of the proposed approach is demonstrated by applying it to fitting the human model to monocular images with real-world test data.