Workplace violence is a serious issue in health care with international surveys revealing disproportionate involvement in certain professional groups, for example, ambulance staff, nurse and student nurses, or settings, for example, mental health and learning disability, elderly care, and A&E units. Staff training is widely advocated as the appropriate organisational response but there are relatively few published evaluations, and so much remains unknown about training effects or effectiveness. Many published studies are flawed by use of small samples, poor control of extraneous, organisational variables, absence of pre-test or follow-up data, limited range of measures, and weak statistical analysis. This study examined an existing training programme for student nurses whilst attempting to avoid the limitations identified above. The effects of training on a number of learning domains, for example, knowledge, self confidence, beliefs and attitudes, and self –assessed skills was investigated using a repeated measures, variable baseline research design, in conjunction with a model of learning. The likelihood of student nurses involvement in violent incidents, and the power/ease of use of different change evaluation methods were also investigated. Repeated administration of a purpose -designed questionnaire at four time points to three consecutive cohorts of student nurses [N=243] provided information about pre-training stability, possible changes on immediate training completion, and at three-months follow-up, after two clinical placements. Statistical analysis revealed the Unit to have generally desirable effects on learning domains that were still detectable at three-month follow-up. It also highlighted differential involvement in violent incidents based on placement type, and important differences between evaluation methods in terms of ease of use.