Abstract. In order to improve the prediction of aerodynamic forces on a wind turbine rotor by the blade element momentum method (BEM), airfoil coefficients are extracted from steady-state 3D RANS simulations of a rotor and then applied for steady-state simulations in a BEM code. The extraction is accomplished by using either averaging of velocities in annular sections, or an inverse BEM approach for determination of the local induction factors in the rotor plane. In this way, 3D rotor polars are obtained which are able to capture the rotational augmentation at the inner part of the blade as well as the load reduction by 3D effects close to the blade tip. When using these 3D rotor polars, the radial force distribution from BEM is very close to the RANS result for a variety of load cases, whereas the deviation is often large with 2D airfoil coefficients. However, it is important that the polar extraction is completely consistent the the BEM code in which the polars are supposed to be used. The 3D rotor polars are shown to depend on the blade pitch angle. In addition, the accuracy of the slope of the 3D rotor polars and their feasibility for application in unsteady simulations is assessed by a quasi-steady comparison. This work is an updated and expanded version of a contribution to "The Science of Making Torque from Wind" (TORQUE) 2016 in Munich.