Background: The effectiveness of community-based programmes to treat children suffering from Severe-Acute Malnutrition (SAM) is well documented following their implementation in emergency situations. However, little evidence exists to understand whether these programmes when implemented in non-emergency situation can also achieve successful outcomes. It is only recently (2007) that the community-based approach was approved by the WHO, and adopted by governments of developing countries to be implemented as part of routine health care services. Since then little or no organised research has explored whether the approach has achieved success or not in non-emergency context. It is for this reasons that this research was designed to evaluate the effectiveness of the implementation of the approach in a non-emergency context in Ghana. The aim was to understand the potential impact of community-based programme to treat children suffering from acute malnutrition, as well as identify factors within the Ghanaian context that could potentially influence community-based programmes effectiveness. Method: we adopted a realist mixed method approach to evaluate the effectiveness of the community-based SAM management programme in Ghana. This approach incorporates quantitative and qualitative methods to understand what works, how and why. Quantitative methods were used to retrospectively collect and analyse data of children who attended the programme between July 2010 and January 2011 in Upper East Region. STATA version 11 was used to analyse the data to estimate the proportion of children who recovered, died and defaulted. Multiple backwards logistic regression was used to assess possible predictors of the outcomes achieved. Contextual factors (barriers and facilitators of programme impact) were assessed using qualitative investigation approach. Using this method, semi-structured interviews and focus groups were conducted with stakeholders, including services providers and beneficiaries of the programme to collect in-depth data. The data were transcribed and analysed using a framework approach. Findings: A total of 525 children were enrolled in the programme within the study period. However, only 488 children data were analysed because incomplete data were recorded for 37 children. Of 488 children data analysed, 30% (n=146) did not complete the programme. Of the non-completers, 28% (n=138) has dropped out, and 1.6% (n=8) died. 72% (n=350) of the children recovered from SAM. This proportion include 2% (n=8) of the children who dropped out. Recovery and mortality rates compared favourably with the Sphere acceptable indicators1. Default rate however was comparatively higher and nearly double the Sphere acceptable indicators. The regression analysis suggests no correlation between the variables tested and higher dropouts. However, the qualitative findings, suggest that contextual factors: distance to treatment centre, lack of support from husbands, busy schedule of women, community believes in traditional medicine as a cure for malnutrition, and health system factors: lack of incentives for health workers and volunteers, and inadequate health workers leading to insufficient monitoring of the programme and lack of counselling of caregivers, were potential reasons for the higher dropouts rates recorded. Conclusion: The findings suggest that community-based programme can achieve success when mainstreamed within routine health services and implemented in non-emergency context. However, success in this context can be diluted by higher default, if factors causing this are not identified and addressed. Health systems strengthening could also be a sure way to ensure success, as well as improve impact of community-based programmes in Ghana.