Abstract: The autocorrelation method considers the ionospheric characteristic of interest as a realization of a random process, modeled as the sum of a periodical component and a random Gaussian process, assuming stationarity during the period of interest. Then the process is defined by its autocorrelation function. Based on the autocorrelation analysis theory, a proper steady ionospheric parameter is uesd and the corresponding autocorrelation model which is positive definite is also established in this paper. On this basis, an interpolation method for ionospheric missing data is presented. This method improves the interpolation accuracy of the method given by Muhtarov and Kutiev in 1999. It generally decreases the error by 1~2 percent, and even by nearly nine percent sometimes. Therefore it largely advances the effect of interpolation for ionospheric historical missing data. Furthermore, the variation rules of the interpolation error along with seasons, solar activites and latitudes are analysed.