In social work, practitioners are often faced with situations in which they have to choose an intervention. A fundamental ethical principle of social work practice is to minimize the risk of adverse effects caused by social services. To adhere to this principle, practitioners must be aware of the possible positive and negative effects of potential services. There are hundreds of interventions currently in use in social work. Although there are a growing number of controlled trials on social work interventions, there are only a few interventions based on research that has the highest degree of certainty, making it difficult to know how and in what way these interventions can support a person in need of assistance. This article, based in part on the experience we gained training practitioners working in Swedish social services, presents a tentative model for assessing the best available evidence comparing interventions when scientific evidence is sparse.