Aiming at the shortcomings of low precision and poor performance in the tracking process of wireless sensor networks (WSNs), a multi-rate tracking algorithm based on unscented Kalman filter (UKF) and covariance intersection (CI) fusion was proposed for hierarchical multi-cluster WSNs. The sensors were divided into several clusters. The sensors in the same cluster can use different sampling and transmission rates to collect and transmit the target data. Firstly, UKF was used to process the data collected by sensor nodes and generate local estimation. Then, the CI fusion algorithm was used to form the fusion estimation from the collected local estimates. By defining an additional weight factor, we can define a stricter limitation for the uncertainty of the real covariance. The simulation results show the effectiveness of the method. The accuracy of multi-rate hierarchical fusion estimation is higher and the effect is more obvious.