This article describes the implementa tion of the Professional Review Action Group (PRAG) model for reviewing cases of disrupted family reunification. The PRAG model provides for care re views, periodic reporting of review find ings (a feedback mechanism for the agency and community), and recom mendations for and documentation of corrective actions. In the first 62 cases reviewed, both family and service deliv ery problems were found to contribute directly to placement reentry. The most frequent contributor to placement reen try was nonresolution of parent prob lem that precipitated placement. Mul tiple service delivery system problems, including high caseload size, staff turn over, and insufficient regulation of re unification practice, interacted with the serious nature of families' problems to reduce the chances for successful reuni fication. The authors describe correc tive actions implemented and conclude that the PRAG model provides an effec tive tool for understanding and ad dressing unwanted policy, program, and practice outcomes. IN 1987, the U.S. Department of Health and Human ervices iden tified unsuccessful family reunifi cation as a child welfare system out come failure, citing national figures regarding the high proportion (29 to 33 percent) of children reentering the child welfare placement system (Fed eral Register, 1987). The agency as serted that when a former foster child reappears at the agency the system has the responsibility to examine in sufficient detail not only the reason for deciding again to place the child in foster care but the factors that led to this traumatic repetition (Gersh n son, 1987, p. 1). To that end, the e partment funded a three-year project to pilot the Professional Review Ac tion Group (PRAG) model for review ing cases of disrupted family reunifi cation in an eight-county area in Indiana. This service evaluation and corrective action model provides for case reviews, periodic reporting of review findings (a feedback mecha nism for the agency and community), and recommendations for and docu mentation of corrective actions (Fig ure 1) (Hess & Folaron, 1992).