The recommended approach helps people interested in measuring corrections performance define their measurement needs and develop a performance measurement system appropriate to those needs. The framework for measuring corrections performance resulted from (1) studying what correctional agencies do and who the major actors are, (2) analyzing problems and issues confronting measurement, and (3) constructing and interpreting measures for the five major program areas -- jails, prisons, probation, parole, and community-based programs. The first step in deciding what to measure is identifying the questions that people want answered about a program's performance. Questions most frequently raised about performance address efficiency, cost effectiveness, equity, service quality, unmet needs, and conformance with government policies. The designer should carefully consider who will use the performance information in order to collect information most beneficial to that entity. Natural constitutents for performance information about corrections programs include researchers, planners, budgeters, public interest groups, funding agencies, legislators, and chief executives, as well as corrections agency heads and program managers. After deciding what to measure and identifying the performance concepts, the designer should think about ways of measuring each concept, research the corrections evaluation and measurement literature, and compile a list of potential measures. Decisions about which potential measures to accept or reject may be aided by applying a uniform set of criteria to evaluate each measure. Criteria frequently used for this evaluation focus on one of four areas: technical adequacy, practicality, utility from a general perspective, and utility from the perspective of the specific use intended for the measure. The report summarizes factors that affect how measurements are interpreted, examines bases for comparing measurements, suggests various ways of constructing performance measures, and presents a way of using statistical models to estimate program effects. Models are developed for the extent and timing of postrelease criminal activity and the wages of individuals who have participated in a corrections program. Footnotes and approximately 30 references are provided.