Prevention is a “hard” problem. We cannot just give out medicines to change behavior, and simple approaches on their own are unlikely to be sufficient. Lifestyle behaviours most relevant for health, such as eating choices or physical activity, cannot be changed just with a single intervention, however “effective” it may be.This implies that effective strategies for prevention of most common risk behaviours must be based on a broad spectrum of interventions, targeted to both environment and individuals, and including structural changes and changes in social norms. For example, health education interventions to prevent obesity are likely to have maximal effectiveness in environments where unhealthy foods have high level of taxation, bike lanes are widespread, and restaurants and catering companies highlight healthy food choices. Likewise, school-based interventions to prevent smoking onset should be provided together with the implementation of school policies against tobacco, delivery of tobacco-free homes programmes, smoking bans in public places, and restriction of smoking in the media and film. In other words, changing health-related behaviours requires complex and synergistic strategies. While there is sufficient knowledge about the effects of some health education programmes and school-based interventions, our understanding of how interventions brought at the level of the environment might work is poor. In order to fill this gap, the scientific evaluation of environmental interventions, as well as of complex community interventions should become a priority. But this raises several methodological questions, from the task of disentangling the effect of single components to that of summarising and de-contextualising results.