The invention discloses a rural power grid intelligent level evaluation method based on a neural network, and the method comprises the following steps: (1) building a rural power grid comprehensive evaluation index system containing intelligent indexes, which comprises a first-level index and a second-level index corresponding to the first-level index; (2) calculating the weight by using an entropy weight method; (3) calculating the weight by using an analytic hierarchy process; (4) calculating the weight by using a neural network method; and (5) comparing the result of the entropy weight method weight obtained in the step (2) and the result of the analytic hierarchy process weight obtained in the step (3) with the weight method obtained in the step (4), and verifying the feasibility of the new method. The invention aims to provide a novel rural power grid evaluation index system, and based on a new evaluation method, the intelligent level of the rural power grid is analyzed more scientifically and more reasonably.