The objective of this study was to quantify the accuracy with which two-dimensional ice accretion simulations can be used on a 1/4-scale airfoil model at low Reynolds number to simulate the aerodynamics of ice accreted on a full-scale airfoil model at high Reynolds number. Six full scale ice accretion castings, representing ice roughness, streamwise ice, horn ice, and spanwiseridge ice, were previously tested in the ONERA F1 wind tunnel at Re = 12.0 x 10⁶ and M = 0.20. These castings were digitized and measured to create sub-scale 2-D smooth and simple-geometry simulations, which were tested in the University of Illinois 3 x 4 ft. wind tunnel at Re = 1.8 x 10⁶ and M = 0.18. C_l, C_m, and C_d for each sub-scale simulation was compared with the corresponding full-scale casting. Geometrically-scaled simulations of the horn-ice and spanwise-ridge ice castings modeled C_l,max to within 2% and C_d,min to within 15%. Geometrically-scaled simulations of the ice roughness and streamwise ice tended to have conservative C_l,max and C_d. The aerodynamic performance of simulations of these types of accretion was found to be sensitive to roughness height and concentration. Scaled roughness heights smaller than those found on the casting were necessary to improve simulation accuracy, resulting in C_l,max and C_d,min within 3% and 5% of the casting, respectively. Removing highly three-dimensional features, such as ice feathers, that became two-dimensional in the creation of 2-D smooth simulations was found to improve simulation aerodynamic fidelity.