A method of estimating the parameters of the Rasch Model removing the effect of random guessing is presented. The procedure is an application of the ARRG (Abilities Removing Random Guessing) model recently developed for two parameter latent trait models. Under the Rasch model ARRG provides for estimation of abilities, removing the effects of random guessing, without requiring the use of a computer. Monte Carlo simulations are employed to examine the accuracy of the resulting ability estimates. (Author/RC) (20pp.)