Transient Electromagnetics Method (TEM) field observation data (induced electromotive forces transients process with time) usually contain random fluctuation disturbance.The traditional method to suppress the random noise is to use of" multiple stacking" observation, take the mean value as the observation results, hundreds times of stacking; this method has a certain blanket jamming effect, but it costs of increasing the (supply) energy consumption and reducing the observation efficiency, and is not a good one.In recent years, in order to improve the anti-interference ability of observation data, advanced instruments has been using intensive sampling for transient, recording thousands of transient sampling points.For the sampling data of this kind of instrument, five points or seven points of traditional data smooth processing algorithm cant fully take advantage of the dense and numerous sampling, and the effect is not ideal to suppress the random noise.Therefore, we studied the polynomial fitting to the observed data, and the effect of suppressing random disturbance was well.This paper will introduce the polynomial fitting algorithm, program and the effect of suppressing random interference.