The initiative is structured to follow the stages of implementation science (Fixsen, et al., 2005) which begins with an exploration of risk and protective factors related to the problem (long term foster care) and the target population. Each project was charged with analyzing existing data to identify target populations at the highest risk of poor permanency outcomes to guide organizational decision-making for selecting interventions with the highest degree of evidence for addressing these barriers. Despite the need for social services agencies to use research to guide organizational decision-making, prior research suggests that political and organizational factors may impede true collaboration and the utilization of research results by agency leadership to guide decision-making (Anderson, 2001). Child welfare agencies, in particular, are not typically structured to analyze and use data to drive decision-making about policy and practice reforms (Jack et al., 2010). The projects represented in this symposium employed collaborative relationships between public and private agencies and Schools of Social Work to conduct studies on three target populations at risk for long term foster care: (1) African American and American Indian children; (2) unsafe children due to maltreatment; and (3) children with serious emotional and behavioral problems. Paper 1 uses administrative data to understand barriers for African American and American Indian ethnic groups. Paper 2 uses data from case reviews and administrative data systems to identify the most important family and service characteristics for children who entered care because of maltreatment and safety concerns. Paper 3 explores the trajectory of children identified with serious emotional and behavioral disorders based on data from case reviews and administrative data systems. Data for all three papers were explored through survival analysis procedures to model time to permanency with cohorts of children who entered care in those service jurisdictions. Presenters will discuss the methods used to engage agency partners in defining the scope of data exploration and to consider the results of analyses in their eventual selection of practice/policy strategies that will be tested with these three target populations. Participants will be engaged to consider the benefits of university-agency partnerships for using implementation science to drive system and practice reforms. Implications for future research, policy, and practice will be discussed.