Many reports have indicated that, in contrast to waterflooding, oil–water relative permeabilities in polymer flooding are influenced by the polymer properties, such as adsorption and viscoelasticity. Long-term injected polymer solutions flush the reservoir porous media causing wettability alteration and residual oil reduction, which is also an important factor for oil–water relative permeabilities. Therefore, the comprehensive effects of polymer properties and dynamic flushing on two-phase percolation behavior have been the focus of research. In this study, polymer flow experiments, including pre-shearing, relative permeability, and long-term erosion tests, were first conducted. Then, a flushing-interpolation model, including polymer concentration interpolation and surface flux characterization, was developed to dynamically describe the two-phase flow characteristics. Based on the experimental results and the presented dynamic model, polymer-flooding simulations were conducted to analyze the performance of the well and the distribution of the remaining oil. The results indicate that flushing reduces the residual oil saturation, and the viscoelasticity effect further broadens the predominant flushing channels horizontally; both effects lead to higher cumulative oil production and a lower water-cut. Additionally, including the flushing effect in the polymer-flooding model leads to a more reliable numerical prediction of the remaining oil distribution. To account for the comprehensive effects of viscoelasticity and erosion in field applications, water-cut history matching was performed for the studied well group, and the error decreased to 4.39%. Finally, the optimal polymer concentration for high-concentration polymer-flooding programs was determined to be 1450 mg/L.