In 2002 Helena Kraemer and colleagues published an important article on the analysis of clinical trials in mental health, which advocated a planned focus on mechanisms to investigate the processes behind treatment effects. Kraemer et al. considered not only new approaches to mediation analysis, but also a theoretical approach to factors, both pre-treatment and during treatment, that might moderate this mediation. Trials should not just be about whether a treatment ‘worked’, but how it worked; with the results informing modification of the intervention for the next trial by discarding aspects that were not effective and reinforcing aspects that were – an iterative procedure towards greater effectiveness. Can we enjoy similar ambitions for complex interventions within mental health? It is not so long ago when the received wisdom within the clinical and much of the research community was that it was simply impossible in practice to mount randomised controlled trials relevant to the kind of psychosocial interventions we use in child and adolescent mental health (CAMHS). How different the situation is now, with burgeoning interest in a systematic evidence base for psychological treatment and the possibilities for unexpected advances (as well as unexpected harms). Nevertheless it is probably still fair to say that the systematic use of process and mechanism study within trials in our field is the exception rather than the rule. What are the possibilities and implications for our field?