Rudner (2001, 2005) proposed a method for evaluating classification accuracy in tests based on item response theory (IRT). In this paper, a latent distribution method is developed. For comparison, both methods are applied to a set of real data from a state test. While the latent distribution method relaxes several of the assumptions needed to apply Rudner’s method, both approaches yield extremely comparable results. A simplified approach for applying Rudner’s method and a short SPSS routine are presented.