Abstract An analytical method for predicting hydraulic performance of a jet pump installation was proposed by updating the existing graphic method and validated experimentally. Based on the measured dimensionless head-flow rate curve of a jet pump and the head-flow rate curve of a centrifugal pump, the hydraulic performance curves of the jet pump installation were predicted at various dimensionless flow rates and compared with those measured at 4.5 m and 9.0 m suction heights, respectively. It turned out that that the error in the analytical method is overall lower than in the graphic method, as a result, the dependence of the jet pump head on its dimensionless flow rate has been even better represented by the former. Note that the error at a few operating points still is notably large, obviously, an update on the method is need. In do so, a correction factor, i.e. the ratio of a predicted head to the tested head was used to correct the jet pump head Hc. Then influences of suction height, area ratio, nozzle diameter, flow rate ratio, distance between throat and nozzle, length of throat and pump rotational speed on Hc were identified by employing Plackett-Burman experimental design method. The critical factors affecting the correction factor was found to be the area ratio and flow rate ratio. Finally, an empirical formula for the correction factor was best fitted(correlation coefficient 0.99)by using genetic algorithm in terms of flow rate q and area ratio m. The predicted jet pump heads were contrasted with those measured at 3.5 m and 8.0 m suction heights. It was shown that the errors in the head are less than 5% at each working point. Thus, the heads predicted are subject to a high reliability.