Fault identification of wind turbine drivetrain is the key for wind farms to make appropriate maintenance strategies to reduce the downtime and maintenance cost, and also one of the highly discussed issues and difficulties in recent research. Gravitational search algorithm was applied in the optimization of the initial weights and thresholds of BP neural network for the first time. Therefore, a fault identification method using BP neural network based on gravitational search algorithm was proposed and applied in wind turbine drivetrain. Tests showed that the presented method could precisely identify three typical wind turbine drivetrain faults, which were gear wear, tooth breaking and bearing looseness respectively, with higher average accuracy than BP neural network, so the effectiveness of the proposed method is verified.