This paper deals with the processing of topology errors in power system state estimation. An algorithm is proposed for determining a correlation index between symptoms of anomalies in the estimation process and measurements related to the network elements suspect of being misconfigured. A method based on such an algorithm enables one to recognize the occurrence of topology errors in the real-time modeling process and to identify the misconfigurations.The proposed method has been tested on three different power systems, including a real istic network which is part of the Southern Brazil interconnected system, for various types of topological errors (inclusion, exclusion and bus-split errors).