Abstract: In many process of data analysis, especially in data reconciliation, it is required to know the second order character of data. It is usual to estimate the error variance in the existing method of second order character. These algorithms have common localization limitation, i.e., assume that there don′t exist gross error in samples and samples′ distribution is normal, which connot be satisfied in many real applications. Aiming at the above shortcomings, this paper proposed a new estimation algorithm for data′s second order character. In the proposed the probable variance is utilized to estimate data′s second order character by basing on the relationship theory of series. Both theory analysis and simulation results validate effectiveness of the new algorithm.