This paper examines the development and evaluation of criterion-referenced measures, and elaborates on the distinction between them and norm-referenced measures. The concept of sensitivity is introduced as an appropriate method for evaluating such measures, and a sensitivity index is proposed. The traditional model for the response of a subject to a measure is extended to conform to the typical objective-based measurement situation, and suggestions for using this response model for evaluating the sensitivity of a measure are offered. Alternate versions of the response model, designed to account for a variety of measurement situations, are presented, along with a discussion of the role of item selection and the effect of guessing in the evaluation of the measures. Data from a variety of sources (both empirical and simulated), which was used to study the effects of varying test parameters on the sensitivity of the measures, and the implications of the results of this data are discussed. General considerations for test development, possible uses of the proposed sensitivity index, and recommendations for further research conclude the report. (AG)