The invention discloses a robustness real-time scheduling method in consideration of the estimation error of an injection transfer distribution factor. The method comprises the steps of firstly, establishing a probability estimation model for an injection transfer distribution factor based on the Bayesian linear regression theory by utilizing real-time measurement data; and secondly, solving to obtain an error interval of the injection transfer distribution factor. Furthermore, the method is targeted to the operating economy of the power grid, and a real-time scheduling model in consideration of the node injection power disturbance and the estimation error of the injection transfer distribution factor at the same time is constructed. Meanwhile, based on the Soyster robustness optimization method, the solving algorithm of the model is provided. According to the technical scheme of the invention, the effectiveness of the method is verified through the test, the calculation and the analysis of a simple 6-node system, an IEEE 118 node system and an IEEE 300 node system.