Discharge measurement of irrigation canal is important for planning and managing water resource. Stage-discharge relation model is used for discharge measurement, which is regarded as a traditional soft measurement method. In order to fit out an accurate stage-discharge curve, enough sample data of water stage and discharge are required. Usually, the process of extracting sample data is labor-intensive and time-consuming. Aiming at improving the accuracy and efficiency, this paper puts forward an optimization measurement method for stage-discharge relation model. Firstly, repeated measurement is implemented separately at several steady flow status. For each steady flow status, filtered sample data of water stage and discharge are calculated through filtering method. Then, the filtered sample data are used for searching the solutions for stage-discharge relation model through genetic algorithm. Lastly, Kalman filter is used to calculate the optimal solution for the stage-discharge relation model. By this way, the demand quantity of sample data decreases significantly. Some experiments are conducted, and the results indicate that the performance of optimization measurement method is superior in accuracy and efficiency to that of traditional method. The stage-discharge relation model built by the proposed method can serves for engineering implementation in discharge measurement.