Abstract: The whole society has higher requirements for power supply reliability, which requires the distribution network to have good self-healing properties. In terms of self-healing evaluation of distribution network, the existing researches often neglect the evaluation of the hidden trouble of self-elimination of the distribution network in the non-fault state. In order to comprehensively evaluate the self-healing level of distribution network, not only considers the self-healing situation of distribution network after fault, but also considers the risk assessment of distribution network under non fault state. In order to quantitatively evaluate the risk faced by distribution network in non fault state, based on the risk theory, three quantitative indicators are put forward: line overload risk, voltage overrun risk and load loss risk; in terms of self-healing evaluation of distribution network after fault, it puts forward self-healing speed, self-healing rate, self-healing control operation complexity, self-healing control operation complexity, self-healing rate and self-healing control operation complexity. There are eight quantitative indexes, including voltage/current recovery rate, load shedding rate, self-healing input cost to effect ratio, average outage frequency and average outage duration. According to the characteristics of the above index system, the improved analytic hierarchy process (AHP) based on the extended weighted average operator is used to comprehensively evaluate the self-healing of distribution network. Finally, a simulation example is given to verify the effectiveness of the proposed method.