In this work, the surface roughness (SR), surface hardness (SH), and the thickness of the affected layer (TL) of the multi-roller flat burnishing process are optimized.The parameter inputs are the tool rotational speed (S), burnishing depth (D), and feed rate (f). The flat burnishing tool having three rollers was utilized to facilitate burnishing trials. The Kriging models of performances are proposed regarding inputs.The CRITIC method and Crow Search Algorithm (CSA) were employed to select weights and optimality. The optimizing outcomes indicated that the optimal values of the S, f, and D were 912 rpm, 150 mm/min, and 0.12 mm, respectively. The improvements in the SR, SH, and TL were 33.3%, 26.9%, and 48.6%, respectively. The SR was primarily influenced by the f, followed by the D and S, respectively. The SH and TL were primarily influenced by the D, followed by the S and f, respectively. The optimal data could be applied to the practical multi-roller burnishing process to improve surface properties for flat surfaces. The Kriging models and CSA could be efficiently utilized to solve complex issues for burnishing operations and other machining processes.