OBJECTIVE: Animal-assisted therapy (AAT), most frequently used with dogs, is being used increasingly as an adjunctive alternative treatment for psychiatric patients. AAT with larger animals, such as horses, may have unique benefits. In this randomized controlled study, equine and canine forms of AAT were compared with standard treatments for hospitalized psychiatric patients to determine AAT effects on violent behavior and related measures. METHODS: The study included 90 patients with recent in-hospital violent behavior or highly regressed behavior. Hospitalization at the 500-bed state psychiatric hospital was two months or longer (mean 5.4 years). Participants were randomly selected to receive ten weekly group therapy sessions of standardized equine-assisted psychotherapy (EAP), canine-assisted psychotherapy (CAP), enhanced social skills psychotherapy, or regular hospital care. Participants' mean age was 44, 37% were female, 76% had diagnoses of schizophrenia or schizoaffective disorder, and 56% had been committed involuntarily for civil or forensic reasons. Violence-related incident reports filed by staff in the three months after study intake were compared with reports two months preintake. RESULTS: Interventions were well tolerated. Analyses revealed an intervention group effect (F=3.00, df=3 and 86, p=.035); post hoc tests showed specific benefits of EAP (p<.05). Similar AAT effects were found for the incidence of 1:1 clinical observation (F=2.70, df=3 and 86, p=.051); post hoc tests suggested benefits of CAP (p=.058) as well as EAP (p=.082). Covariance analyses indicated that staff can predict which patients are likely to benefit from EAP (p=.01). CONCLUSIONS: AAT, and perhaps EAP uniquely, may be an effective therapeutic modality for long-term psychiatric patients at risk of violence.