Polymer injectivity is an important factor for evaluating the project economics of chemical ?ood, which is highly related to the polymer viscosity. Because the ?ow rate varies rapidly near injectors and signi?cantly changes the polymer viscosity due to the non-Newtonian rheological behavior, the polymer viscosity near the wellbore is dif?cult to estimate accurately with the practical gridblock size in reservoir simulation. To reduce the impact of polymer rheology upon chemical EOR simulations, we used an ef?cient multilevel local grid re?nement (LGR) method that provides a higher resolution of the ?ows in the near-wellbore region. An ef?cient numerical scheme was proposed to accurately solve the pressure equation and concentration equations on the multilevel grid for both homogeneous and heterogeneous reservoir cases. The block list and connections of the multilevel grid are generated via an ef?cient and extensible algorithm. Field case simulation results indicate that the proposed LGR is consistent with the analytical injectivity model and achieves the closest results to the full grid re?nement, which considerably improves the accuracy of solutions compared with the original grid. In addition, the method was validated by comparing it with the LGR module of CMG_STARS. Besides polymer injectivity calculations, the LGR method is applicable for other problems in need of near-wellbore treatment, such as fractures near wells.