Nursing assistants (NAs) working in long-term care have the highest incidence of workplace assault among all workers in the United States. The objective of this study was to test the effectiveness of a violence-prevention intervention, on the basis of Social Cognitive Theory to increase knowledge, self-efficacy, and skills, and to decrease assaults. Investigators conducted a quasi-experimental study with 138 NAs in 3 intervention and 3 comparison homes. A baseline questionnaire was used to obtain information on demographics, employment, and violence experience. At pre-, post- and 6 months after the intervention, all participants completed the State Trait Anger Inventory and the Knowledge and Self-Efficacy Survey, carried an Assault log for 80 hr of work, and participated in a simulation exercise to assess violence-prevention skills. Tabulations, analysis of variance, and Poisson regression were used to analyze the data. The intervention participants showed significant increases in knowledge, self-efficacy, and violence-prevention skills. Although the intervention had no significant main effect on the incidence of assaults, there was an interaction effect between the intervention and the number of preintervention assaults. The intervention had a significant effect on those NAs who had fewer than 6 assaults preintervention (p < .001) and no significant effect on those who had more than 7 assaults on preintervention. There were significant relationships between assaults and the following covariates: age, state anger, and the number of residents assigned. Although the incidence of violence cannot be eliminated, it can be decreased and it should never be tolerated or accepted as "part of the job" because to do so devalues the NA.