Realizing precision machining of pre-mixed abrasive water jet is an important issue that needs to be solved, and physical model analysis, experimental research, establishment of predictive mathematical models, and experimental verification are relatively mature solutions. Based on dimensional analysis, the key influencing parameters on the cutting surface roughness Ra of pre-mixed abrasive water jet cutting were determined: working pressure of water jet P, feed rate v, target distance h, and nozzle outlet diameter d. A full factor cutting experiment L81 (34) was developed for Q235 carbon structural steel based on four key influencing parameters, and the cutting experiment was completed using a pre-mixed abrasive water jet cutting system. Based on the analysis of polar variance, the significance relationships of four key influencing parameters were obtained, with d>v>P>h. Analyze the experimental data to obtain the optimal parameter combination: P=28 MPa, v=50 mm/min, h=4 mm, d=1.14 mm. The minimum Ra under this parameter combination is 0.760 µm. Based on the multivariate nonlinear regression analysis method, a mathematical model for predicting the cutting surface roughness of Q235 carbon structural steel was constructed, with an average relative error of 3.633%. The mathematical model for predicting the cutting surface roughness of Q235 carbon structural steel constructed in this study has good reliability and has a good predictive effect on the cutting surface roughness.