Temperature control has received widely attention in open-cathode PEM fuel cell system, due to the importance of stack temperature to the output performance of the stack. The open-cathode PEM fuel cell is such a nonlinear, time varying and uncertain system which increases the difficulty of establishing a control-oriented mechanistic model. In this paper, the on-line T-S fuzzy identification method is chosen to identify the open-cathode PEM fuel cell temperature dynamic model. The on-line T-S fuzzy identification method has the advantages of low computational cost and less requirements of the model parameters. It becomes one of the most promising methods to identify the system dynamic model. To appraise the applicability, efficiency, and robustness of the proposed on-line identification method, a real open-cathode PEM fuel cell platform was set up. Experimental results show that the established on-line T-S fuzzy identification model is sufficiently accurate to follow the temperature response of the stack, and can be easily utilized to model-based temperature control system design.