Aiming at the difficulty of determining the exact topology of distribution network during long-term reliability evaluation, a weighted cloud prediction model is proposed to predict the reliability of distribution network in long-term planning. Firstly, the digital features of cloud model of the influencing factors are obtained from the statistical historical data by an improved backward cloud generator. Secondly, based on the above digital features, the average weighted cloud prediction model are set up where the weight of each influencing factor is calculated by analytic hierarchy process according to the sensitivity of the influencing factors on reliability. Finally, the effectiveness and feasibility of the proposed method is demonstrated by comparison with the traditional cloud prediction method and the least square support vector machine method from the indices of maximum, minimum and average error.