The possibility of harm from mental health provision, and in particular harm from youth mental health provision, has been largely overlooked. We contend that if we continue to assume youth mental health services can do no harm, and all that is needed is more services, we continue to risk the possibility that the safety of children and young people is unintentionally compromised. We propose a three level framework for considering harm from youth mental health provision (1. ineffective engagement, 2. ineffective practice and 3. adverse events) and suggest how this framework could be used to support quality improvement in services.