A fusion algorithm for millimeter wave (MMW) radar and infrared (IR) sensor is discussed in this paper. Modified extend Kalman filter (MEKF) is simple yet very effective in accounting for the measurement of non-linearity. On condition of combining MEKF with sequential filter, the optimum state of maneuvering target is obtained. The target is tracked with MMW radar using MEKF, and then the filtering results are fused with information from IR sensor through sequential filter. In this way, the global state is updated at the fusion center. Simulation shows that the novel method can significantly improve the state estimation precision of composite guidance system and strengthens the adaptability of the tracking system to environmental changes.