Introduction : Coordination between levels of care is not facilitated in the Belgian health system. Indeed, patients have uninhibited access to every level of care, there is no gatekeeping system, and few structural coordination between levels of care. On one hand, the occurrence of more complex care situations in the ambulatory setting is enhancing the need for coordination while on the other hand, hospitals face financial constraints to provide care in the community. Implementation : The aim of the research was to organize coordination between levels of care at the local level, in an urban setting. We used the “Local Health System” model (LHS) that aims at integrating hospital and primary care activities for a defined population at the local level. We chose the participatory action research (PAR) methodology and its spiral plan-do-study-act cycles, to ensure the participation and implementation of results by actors of the two levels of care. Key findings : Activities and outputs of the 4-year PAR are presented tough the research is still running. The first step of the research highlighted the interest that representatives of professionals (GP and hospital specialists) have for coordination between levels of care. It also revealed a lack of awareness regarding the role of respective work organizations and the difficulties to act as representatives. A coordination platform between one hospital and local GP organizations was developed with technical and organizational support of researchers. Regular meetings of representatives were organized, without financial incentive or policy constraint. The first and main activities were oriented toward explanations of the specific tasks and competencies of each levels of care. Practical improvements concerning coordination mechanisms were recorded such as transfer of information between levels of care, direct contacts between GP and specialists and definition of each other’s scope of action in routines for complex situations. Improved integration between organizations also improved representation mechanisms and information flow within the organizations themselves. Interest from neighboring hospitals grew and led to the emergence of other coordination platforms between levels of care in the same area, some centered on one hospital and others involving several hospitals around one disease. Highlights : Our interventions filled an operational gap and as such, were supported by local actors, hospital and GP organizations. Some recent health policies were also identified as action-levers. Our PAR suggests also that immaterial incentives and appropriate bottom-up organization may significantly improve local integration of care in the context of fragmented health system. Currently the coordination platforms revolve around one hospital, according to specific organizational constraints of each institution and market-based organization of secondary care. The coordination process remains doctor-centered and single-disease oriented, reproducing the current organization of care and the prevailing approach of chronic disease. The inputs of researchers in the process emphasize the importance of raising participants’ awareness of organizational integration of care and supporting boundary spanners’ competences in integrating levels of care. Conclusion : Our bottom-up approach proved multi-functional and cheap. Although time-consuming, it ensured adoption and sustainability of the process by the actors. If policy support may help sustainability and transferability of LHS within health systems with weak integration between levels of care by fostering functional integration, the importance of voluntary adhesion and strategies design adapted to local characteristics proved essential.