In view of the compound homing problem of a missile under strong counterwork, a radar/IR sequential filtering fusion algorithm which controls radar under intermittent-working state was proposed in radar/IR compound guidance system. In the algorithm, the sequential filtering fusion algorithm based on interacting multiple model and unscented Kalman filter (IMM-UKF) was used for target tracking. Through comparison between the state estimation covariance with the measurement error variance, the algorithm controls radar under intermittent-working state. The simulation results show that, compared with the data compression target tracking, the algorithm has higher tracking precision, and can reasonably control radar radiation on the premise of considerable tracking precision.