Aiming at the problem of the diversity of fault causes and the random uncertainty caused by the increasingly complex distribution network structure, a random fuzzy?based distribution network failure risk assessment method is proposed. Compared with the traditional risk assessment of power outages in distribution networks, this method describes the cumulative probability of failures in a certain area from a data?driven dimension through random fuzzification of parameter models. In response to the increasing power supply reliability expectations of power users, the entropy weight method is used to comprehensively consider factors such as the fault level of the power outage accident, the length of the power outage, the number of sensitive users and the number of affected users, and conduct risk assessment of the distribution network failure. The results show that the probability distribution of the number of failures calculated by using the random fuzzy theory can realize a more scientific and comprehensive analysis of the distribution network failures. The weight value of the risk assessment index determined by the entropy weight method can provide support for formulating reasonable and effective fault repair strategies, formulate response plans for sensitive users and increase the degree of power user ownership.