Abstract: The Adaptive Fractional Lower order Covariance (AFLC) time delay estimation method performs robust under impulsive noise environments, but the values of parameters a and b have effects on estimated precision. According to the non-stationary property of noises, this paper proposes a dynamic parameter estimation method and further proposes a Modified Adaptive Fractional Lower order Covariance (M-AFLC) method. Computer simulation indicates that the iterative parameter estimation method performs well under both stationaiy and non-stationary noise conditions. It also shows that the M-AFLC maintains the merits of the AFLC, and at the same time the proposed method avoids the degradation of AFLC while the restriction unsatisfied.