The invention provides a maximum absolute value inverse proportional robust state estimation method. The method comprises the following steps of A. providing a basic model for maximum absolute value inverse proportional robust state estimation; B. introducing an auxiliary variable for the basic model for maximum absolute value inverse proportional robust state estimation, and changing to obtain an equivalent model for maximum absolute value inverse proportional robust state estimation; and C. solving the equivalent model for maximum absolute value inverse proportional robust state estimation by using a primal-paired interior point method. Example analysis shows that the method can effectively suppress multiple bad data including data with poor consistency in the process of estimation, thereby showing very high robust state estimation performance, having very high calculation efficiency, and being very suitable for actual projects.