Self-resonating cavitation waterjet (SRCW) has rapidly developed and been widely used in cleaning fields due to its non-thermal and environmentally friendly machining. However, its fuzzy relationship between cleaning effect and working parameters is a significant limit to cleaning quality improving and extensive application of SRCW. Addressing this issue, an evaluation method of cleaning effect induced by SRCW was proposed in this paper, which includes two evaluation parameters, cleaning rate (R c , mm 2 /s) and primer damage rate (R d ). Using this method, experiments on the effects of four working parameters (pressure, traverse rate, standoff distance and impact angle) on the cleaning effect were carried out, and the results showed that R c and R d is proportional to pressure, with the decrease of attacking angle or the increase of traverse rate and standoff distance, R d always decreases while R c increases first and then decreases. A prediction model of SRCW cleaning effect considering pressure, traverse rate, standoff distance and impact angle was established with the use of GA-BP neural network. The prediction results indicate that the model has better adaptive ability and high prediction accuracy of over 95 %. And according to the prediction and different optimization objectives, the working parameters of SRCW were obtained. The study provided a new method to evaluate the cleaning effect of SRCW, and the working parameters were predicted and optimized by reliably modeling, which is of great significance to guide the practical application of SRCW.