In order to study the autocorrelation properties of 900MHz radio signal, the time series theory is introduced into statistical analysis. By means of the statistical data from signal acquisition system, the Box-Jenkins method is taken into radio signals analysis. From autocorrelation function calculation, it shows the radio signals exist regularity, nonstationarity and nonseasonal in this period. On the other hand, the results show that the autocorrelation of radio signals can be diminished by difference, but as the order of difference increases, the number of correlated coefficients goes beyond the confidence limits more, which implies the degree of autocorrelation can't be eliminated forever as the times of difference increase.