The concept of the ''critical ionization velocity'' (CIV) of a neutral gas has received much attention since it was proposed by Alfv¿n. In the last several years, numerical simulations have been a useful method of examining the physical mechanism behind CIV. Since these simulations have been run using explicit particle-in-cell (PIC) codes, artificial mass ratios (e.g., mi/me=100) have been employed to reduce the time scales in the simulation and the computational power required for the problem. However, the effect of the use of unphysical mass ratios on the simulation results has not been well discussed. In this study, simulations using artificial and realistic mass ratios are compared. The study employs an implicit PIC code to allow the realistic mass ratio simulations to be performed efficiently. Several numerical aspects of using implicit codes, such as the undesirable damping of upper hybrid heating, are discussed. The results indicate that when scaled appropriately, simulations using physical and unphysical mass ratios provide nearly identical results when the anomalous ionization rate is either much lower than the ion cyclotron frequency (&ngr;ion/&OHgr;i≪1) or greater than the ion cyclotron frequency (&ngr;ion/&OHgr;i>1). In the case of an anomalous ionization rate near the cyclotron frequency, the unphysical mass ratio results cannot be scaled easily to regain the results from the physical mass ratio simulation. ¿ American Geophysical Union 1992