The rapid advancement of computational technology has positioned Computational Fluid Dynamics (CFD) as an effective tool for predicting marine propeller hydrodynamic performance. This study applies a CFD-based numerical method using Reynolds-Averaged Navier–Stokes (RANS) equations to evaluate and compare the accuracy of three turbulence models, namely k–ε, k–ω, and k–ω SST for a Gawn Series marine propeller. CFD simulations were conducted by varying the advance coefficient (J). Verification of the results for each turbulence model was carried out through Grid Independence Study and Grid Convergence Index analysis. The CFD simulation results were then compared with experimental test data using the Root Mean Square Error (RMSE) method, with KT, 10KQ, and efficiency as evaluation parameters. The results demonstrate that the k–ω SST model provides the most consistent and accurate predictions across the entire operating range, with RMSE values of 1.24% for KT, 1.98% for 10KQ, and 1.88% for efficiency. Pressure‑contour visualization from the k–ω SST model shows the smoothest pressure distribution on the blade surface, while pathline visualization reveals the clearest, most consistent, and well‑balanced vortex structures downstream of the propeller, thereby providing a robust and reliable basis for selecting the most suitable turbulence model to improve the accuracy of CFD based marine propeller performance prediction.