This article aims to show the interest of a mixed and participative methodology for the prevention of psychosocial risks at work using a “psycho-socio- organizational” diagnosis conducted in a host institution for elderly people called « The Residence ». Numerous studies have shown how psychosocial risks and well-being at work have affected both the physical and mental health of workers in the medical field, and the DARES’ results indicate that occupations involving care and health sectors are particularly affected by two specific dimensions: job demands (frequent schedule changes, working at night or on Sundays and during holidays) and emotional demands (regular assistance and support to people suffering from physical and / or mental disorders and limited autonomy). The quantitative methodology is based on a self-evaluation questionnaire and three validated scales: the Job Content Questionnaire (JCQ), the Maslach Burn-out Inventory (MBI), and the Hospital Anxiety and Depression Scale (HADS). The qualitative methodology is based on inspection of internal documents of the Residence, semi-structured interviews and observations of work situations. The results indicate that average scores on psychosocial risks for the Residence fall below national averages but vary greatly as a function of social professional categories. The JCQ average of “Psychological demands” of the Residence (which includes working time) is higher than that of the SUMER. The nurses’ scores are the highest. “Emotional requirements” should be understood in relation to the prevalence of verbal and physical violence. Health care assistants, ancillary staff, medical / psychological assistants and nurses are at the forefront of these types of problems. In the discussion, we show how a mixed and participative methodology is part of the process of psychosocial risks prevention highlighting various origins of PSR, and allowing that recommendations be co-constructed through the diagnosis, which is essential for the sustainability of the actions to be implemented. Our intervention research provides a diagnostic model transferable to other structures.