We attempted to cross-validate the Classification of Violence Risk, or COVR (Monahan, Steadman, Appelbaum et al., 2005), in a sample of acute psychiatric patients in the North West of England. The study used a prospective design to follow-up 93 patients assessed using the COVR for 20 weeks post-discharge. Overall, the COVR did not demonstrate significant predictive validity with respect to violence. More than half of patients classified as average risk by the COVR were violent in the follow-up compared to less than 20% of those in the very low risk group, but none of the high-risk participants were violent. In contrast, violence was significantly predicted by key variables, including female gender, violence preceding admission, and personality disorder. There is a need for more cross-validation research of the COVR.