The objective of this study is to determine training which joint is more effective in jump height increase by computer simulation. This will be useful information for coaches and athletes in their training strategies. A planar human model with five rigid segments representing the feet, shanks, thighs, TH (trunk and head), and arms is used. Segments are connected by frictionless revolute joints representing the ball of foot, ankle, knee, hip, and shoulder. Model movement is driven by torque actuators at all joints except for the ball joint. Each joint torque is the product of maximum isometric torque and three variable functions depending on instantaneous joint angle, angular velocity, and activation level, respectively. The model can actively extend and flex these joints by changing the activation level. Jumping movements starting from a balanced initial posture and ending at takeoff are simulated. The objective is to find joint torque activation patterns during ground contact so that jump height is maximized. The simulation is repeated for varying maximum isometric torque of one joint ±20% while keeping other joint strength values unchanged. Although jump height increases/decreases with increasing/decreasing strength of each joint, the influence is different among joints. Similar to previous simulation results, strengthening knee joint to 20% is the most effective way than the same level of strengthening for other joints. In general, for the same amount of percentage increase/decrease in strength, the shoulder is the least effective joint in changing jump height.