Background: There is a high rate of physical violence in populations with intellectual disabilities, and this has been linked to problems for the victim, the assailant, members of staff and services. Despite the clinical significance of this behaviour, few studies have assessed methods of predicting its occurrence. The present study examined clinical and actuarial methods of predicting violence in a forensic intellectual disability hospital. Methods: The sample for the study consisted of people resident in the hospital during a 1-year period (n = 124). Clinical prediction used a risk rating made by members of the person's clinical team, whereas actuarial prediction used the number of violent incidents in the 6-months before the date of the clinical risk assessment. Computerized hospital records of violence in the 6 months after the assessment were used to examine the predictive accuracy of the two methods. Results: The clinical method produced an area under the curve of 0.74 (95% CI: 0.65-0.83) in a receiver-operating characteristic curve analysis. The value for the actuarial method was 0.77 (95% CI: 0.69-0.86). Both methods performed at levels significantly above chance, but no one method was found to be superior to the other. Conclusions: These findings suggest that it is possible to predict who is at risk of violence in forensic populations with intellectual disabilities. Here, the authors discuss the clinical implications of these findings and the clinical application of risk prediction within clinical services.