Abstract: A hybrid particle filter was proposed for the synchronization of symbol and carrier frequency in multipleinput multipleoutput wireless communication systems. It is based on maximization of a posterior probability. In the receiver side, propagation delay and carrier frequency offset are represented by a set of random particles. The important weights of particles are calculated by combining the channel parameters, which are estimated by the Kalman filter. Moreover, sequential importance sample and resample are implemented to update the weights and maintain the variety of particles. The approach does not directly sample on multidimensional channel space to overcome the disadvantage of computation efficiency reduction. Numerical results are presented under a fourtransmit tworeceive wireless communication system. They demonstrate that the estimation accuracy of the approach is high as well as the iterative step is short. The bit error rate performance is comparative to the situation with ideal parameters.