Violence and aggression is acknowledged as a serious issue in the mental health services. The aims are to explore whether de-escalation and physical intervention training is effective in reducing incidents and incident severity on a Psychiatric Intensive Care Unit (PICU) and to consider the cost impact. Poisson regression analysis was used to compare the number and severity of incidents on a PICU before and after de-escalation and restraint training. This study shows no significant differences in the number or severity of incidents before and after training. Objective assessment in the evaluation of interventions to improve the safety of the inpatient services is difficult when data is recorded inconsistently or inaccurately. The severity of incidents needs to be defined more fully to allow accurate measurement of the efficiency of techniques employed to resolve violence. The cost impact of training in the management of violence in relation to the benefits remains unclear in the absence of accurate data being recorded.