MIMOOFDM system channel estimation is a key of coherent demodulation for signal receiver. Aimed at the fact that MIMOOFDM system is faced with a nongauss noise, combined with the improved proposal distribution, an improved hybrid annealed particle filter algorithm is used to estimate the MIMOOFDM channel. On the basis of building the statespace model of system and analyzing the proposal distribution of hybrid annealed, the fuzzy inference system is used to get dynamic hybrid annealed parameters, and the improved hybrid annealed particle filter algorithm is obtained. The improved hybrid annealed particle filter algorithm is used to estimate the channel of the MIMOOFDM system. In this process, the bit error rate (BER), the normalized mean square error (NMSE) and the algorithm complexity of the channel estimation are simulated. The simulation results show that compared with the extended Kalman filter, the particle filter and hybrid annealed particle filter algorithm, the improved hybrid annealed particle filter algorithm can reduce effectively the bite error rate of system in nongauss channel, and at the same time, this can also improve the performance of system by using a small amount of sampled particles.