The measurement information generated by CRT's is designed for use in instructional management systems where classifications of pupils for treatment are to be decided on the basis of minimal data consistent with predetermined limits for the errors of misclassification. The measures obtained are content specific estimates of proficiency useful for the stratification of learning groups on a day-to-day basis if need be. By sampling across items rather than across persons, absolute measures of proficiency are obtained which can be reliably interpreted for nonrandomly selected pupils, the pupils of particular concern. The model is designed for wide variety of applications but retains in the concept of proficiency a simple and useful index for instructional management. The empirical data generated have clear implications for instructional decision-making. Some of the applications of item-sampling theory include the ability to (1) categorize learners into temporary learning groups on the basis of a common requirement for instructional treatment; (2) assess the relative effectiveness of competing instructional treatments; (3) determine, in the case of established instructional segments having predetermined performance standards, which individuals require further prescriptive assistance; and (4) in the case of curriculum development, to indicate hierarchical relations within a content sequence. (Author/DB)