Abstract. In Direct sequence spread spectrum (DSSS) communication system, the auto-correlation of Pseudo-Noise Code is weak while the auto-correlation of Narrow-Band Interference (NBI) is strong, so that the NBI is more predictable than the DSSS signal. The principle of time-domain anti-interference is the auto-correlation difference between DSSS signal and NBI. However, due to the existence of oversampling in the transmitter, the auto-correlation of DSSS signal increases significantly, and the difference of auto-correlation between DSSS signal and NBI will diminish. And then the traditional time-domain adaptive filter will introduce distortion to the DSSS signal in rejection of the NBI. In this paper, an improved time-domain adaptive filtering algorithm is proposed. This algorithm is based on the interpolating transversal filter and the Least-mean-square (LMS) algorithm for updating the weights adaptively. Its anti-interference ability gains about 10dB than the traditional adaptive filter under the situation of SNR/chip is greater than 50dB. Both theoretical analysis and simulations illustrate that the new algorithm has a stronger anti-interference ability and introduces less distortion to the signal under high SNR conditions. Meanwhile, it has similar performance with the traditional adaptive filter under low SNR conditions.