The invention discloses a reactive power loss based electric power system SCADA bad data filtering method, which is characterized in that reactive power loss on a line or a transformer at each time period is solved according to reactive power in SCADA measurement data of two ends of the line or a transformer, and then the mean and the standard deviation of the reactive power losses of N time periods are solved; the confidence interval length of the reactive power loss is defined, a measurement data confidence interval for a corresponding time period of a certain line according to the confidence interval length, and the mean and the standard deviation of the reactive power losses of the N time periods, and measurement data distributed beyond the confidence interval is eliminated. Compared with an existing measurement data eliminating method, the method disclosed by the invention is simple and convenient, does not need to build a complex nonlinear equation, is high in execution speed, can be applied to practical engineering, can perform preprocessing on measurement data, and improves the accuracy of state estimation, flow calculation and the like.