The work aims to propose a material removal model suitable for finishing of aluminum alloy workpiece during the wet spindle-type barrel finishing. The FLUENT two-phase flow module was used to simulate the p?v value under different parameters, and the response surface method was used to fit the multiple regression model of four main parameters and p?v. Based on this, a modified material removal model was established based on the Preston equation to perform processing experiments and investigate the relationship between MRR, Ra after processing and surface topography of the workpiece. P value of the workpiece axial distance from the wall and the drum speed term was <0.05 in the analysis of variance and had significant influence on MRR (the former was negatively correlated with MRR and the latter was positively correlated with MRR). The model was verified by experiment. The average relative error of workpiece’s MRRth and MRRex was only 6.32%, thus proving the validity of the model. With the increase of MRR, Ra after processing firstly decreased and then increased. MRR was 21 mg/h, Ra was 0.816 μm and the knife was still very obvious. When MRR value was 59 mg/h, the lowest value of Ra was 0.472 μm, the knife was completely removed, the surface texture was smooth and the processing effect was the best. When MRR reached the extreme value, Ra was 0.693 μm, the surface texture became rough and uneven, and the processing effect was poor due to excessive processing ability of the abrasive block. A material removal model with process parameters as independent variables is established to provide theoretical basis for selecting parameters; and experimental research shows that there is no positive correlation between processing capacity and effect.