The invention provides a method for predicting the operational reliability of a power distribution network based on an ARIMA model, and the method comprises the steps: predicting the number of times of monthly power failures of a user through building the ARIMA model; enabling an unstable element failure time sequence to be converted into a stable time sequence, and then carrying out the regression of the lagged value of a dependent variable and the present value and lagged value of a random error term to build a user monthly power failure time model; sampling a shutdown point according to the prediction results; considering real-time load operation conditions; building a fault mode impact table based on a TLOC rule and a PLOC rule; calculating the system recovery time for the shutdown of equipment at each time, and finally obtaining a yearly reliability index. The method proposed by the invention guides the planning, design, operation and maintenance of a future power grid effectively and accurately, improves the accuracy of prediction and estimation of the operational reliability of the power distribution network, achieves the stable operation of the power distribution network, reduces the frequency of power failures, and reduces the power failure range.