Background: The MacArthur Violence Risk Assessment Study (MacVRAS) in the USA provided strong evidence to support an actuarial approach in community violence risk assessment. Aims: To examine the predictive accuracy of the MacVRAS measures, in addition to structured professional judgement, in a UK sample of patients discharged from in-patient care in the north-west of England. Method: A prospective study of 112 participants assessed pre-discharge and followed up at 24 weeks post-discharge. Pre-discharge measures were compared with prevalence of violent behaviour to determine predictive validity of risk factors. Results: Historical measures of risk and measures of psychopathy, impulsiveness and anger were highly predictive of community violence. The more dynamic clinical and risk management factors derived from structured professional judgement (rated at discharge) added significant incremental validity to the historical factors in predicting community violence. Conclusions: Although static measures of risk relating to past history and personality make an important contribution to assessment of violence risk, consideration of current dynamic factors relating to illness and risk management significantly improves predictive accuracy.