Numerous commentaries plea for the use of methods that take into account complexity when evaluating health programmes. These commentaries provide compelling arguments on why evaluations of health programmes should not rely exclusively on methods that were designed for making universal claims about whether and to what extent a policy is effective. However, there exist only few concrete examples showing how to take this complexity into account during the evaluation of health programmes. One increasingly popular approach to do so is the realist approach. In this paper, we explain the realist approach for taking into account complexity, discuss how we applied this approach to study the impact of smoke-free school policies and reflect on the practical value of the resultant insights. We hope this case study may inspire fellow scholars to use the realist approach for evaluating health programmes.