The study incorporates a set of inputs and outputs that State and Federal agencies as well as prison managers can use to assess the relative efficiency of prison operations and to set targets for improvement. DEA is a linear programming technique that constructs a frontier of relatively efficient decisionmaking units by optimally allocating weights among the multiple inputs and multiple outputs of the units; units falling behind the frontier are classified as inefficient. This analysis demonstrates the general applicability of DEA to prison administration, and serves as a starting point for comparative evaluations of prisons. The study applied DEA to Michigan prisons to identify inefficient prisons, provide some insight into the source of the inefficiency, and identify potential remedies and target performance levels. The DEA process can provide managers a tool to reduce personal bias in the process of resource allocation and budget decisions while establishing greater credibility with their constituency. Notes, tables, references