The recent generation of anti-tank guided missiles uses the reticle seekers to track the IR signature of the missile. The presence of a modulated jammer in the field of view of such a seeker will break the tracking process resulting in a miss of the guided missile. This paper suggests an IR imaging system immune to IR jamming. The system consists of a thermal camera, image acquisition electronics, frame difference operator, a nearest neighbour classifier, and a Kalman filter. The frame difference operator provides a real time processor for rejecting slowly changing and complex background. The resulting image is modeled as stationary cluttered background with a moving missile. The K.F. is used to steer a window in the image field centered at the predicted missile position. The nearest neighbour classifier provides a suboptimal measure of the current missile position. This system has been simulated on a digital computer. The simulation results show the effectiveness of this system in rejecting both natural and artificial decoys without degrading the required missdistance.