The present invention relates to a method for measuring interference power in a time slot CDMA system, The method includes: performing channel estimation for received signals with channel estimation codes, to obtain the original channel response estimation results hi, i=1 . . . P; predetermining a threshold of number of taps W1, and selecting the channel response estimation results for W1 taps with less power from the original channel response estimation results hi for W1 as the roughly estimated result of the interference power; performing threshold processing on the original channel response estimation results by post-processing against signal-to-noise ratio threshold, by using the roughly estimated result of the interference power and the predetermined signal-to-noise ratio threshold, before processing compensating the possible error generated from rough estimation of the interference power and performing threshold processing with the compensated power threshold, so as to ultimately obtain an accurate measured result of the interference power.