A computer simulation of Bayes' Theorem was conducted in order to determine the probability that an examinee was a master conditional upon his test score. The inputs were: number of mastery states assumed, test length, prior expectation of masters in the examinee population, and conditional probability of a master getting a randomly selected test item correct, and of getting an item incorrect. Classification accuracy was shown to be a function of all of the above parameters for any specified level of mastery (in the criterion-referenced sense). Specific results showed that for some combinations of prior information and test length, no information from the test could force a reversal in the decision rule, or provide classification accuracy within acceptable error bounds...hence, test results would be irrelevant. The vulnerability of a Bayesian model to changes in the prior probabilities was also demonstrated. For example, a 10% change in conditional probability was sufficient to completely reverse a classification rule across all test lengths studied, when the prior probability was held constant. Less drastic shifts occured with changes in the prior probabilities. (Author)