With the comprehensive construction of smart grid, it is possible to evaluate the running state of electricity meters by collecting data of electricity consumption of smart meter terminals. This paper presents a reliability evaluation method of electricity meters based on fault correlation degree of abnormal events. Due to the coupling of abnormal event alarm caused by watt-hour meter fault, abnormal event is divided into single abnormal event and concurrent abnormal event for analysis. For the single abnormal alarm event, the direct relationship between the frequency of abnormal alarm and the failure of the watt-hour meter is analyzed; For combined abnormal alarm events, Apriori algorithm is used to select the frequent combined abnormal alarm events, and then the fault correlation degree of the abnormal events is calculated. At last, the reliability of electricity meters in the same area is evaluated by fault correlation degree of abnormal events. The feasibility of this method is verified by an example.