Abstract: A concept of human body's comfort level is defined by methods of fuzzy multi-objective evaluations. Based on evaluations of the body's comfort level, a rule-based strategy with multiple stages of behavior trials is applied to plan virtual human's behavior. Nash's theory of equilibrium is used to optimize virtual mannequin's working gestures in this strategy. The method can be used to generate sequential simulations of human's behavior for adapting environment and provide synthetical results of ergonomic evaluations.