The effectiveness of a model for determining a minimal cut-off score for criterion-referenced tests was examined. The model, based upon techniques presented originally by Nedelsky and by Angoff, was first used in conjunction with a multiple choice test developed for use in certifying school counselors in Georgia. A "knowledge estimation panel" was used to determine criterion scores; the panel of subject experts estimates the percentages of minimally knowledgeable candidates who could be expected to know (not guess) the correct answers. Nineteen judges on the panel rated 684 items. The resulting criterion scores were then lowered to adjust for standard error and raised to adjust for guessing. The consistency of results was supported by a comparison of parallel test forms and comparable halves of the panel. The panel's estimates of item difficulty were found to be highly correlated with actual difficulty when trial tests were administered. Beta coefficients of agreement also supported the consistency of mastery decisions made. (The legal importance of the models' validity is briefly discussed, and several references to court cases are appended). (GDC)