Abstract: An unbiased update form of Hankel matrix is designed based on both the data form with forgetting factor and the construction of Hankel matrix. On this basis, a new recursive subspace identification algorithm with variable forgetting factor is proposed. The modified method, the fixed forgetting factor method and the traditional variable forgetting factor method are compared from the aspects of the approach degree to the unbiased update Hankel matrix, the performance in tracking system parameters and the distance sensitivity, etc. Simulation results show that the performance of the modified method is better than the others.