The proposed standard for judging proficiency test score reliability requires that the proportion of items passed for each objective assessed be a dependable estimate of the universe score for the domain strata established by the objective. Domain breadth is the focusing issue. Data from a field trial of the Tennessee Proficiency Test are analyzed to determine the dependability of measurements for each objective and the minimum number of items needed to assess each objective reliably. The test reports performance in each of four domains (mathematics, reading, language, and spelling) as a single summative score contrasted with the standard to describe a student as a master or non-master. The reasonableness of criterion-referenced interpretation of scores, the usefulness of empirical approaches to strengthening the power of the interpretations and implications of the findings for the construction of proficiency tests are discussed. (Author/CM)