This study considered the ability of items on the HCR-20 and VRS, structured clinical risk assessments, to predict inpatient violence within the first 6 months of admission to a secure forensic service. A retrospective file review of information available at time of admission was used to complete the Historical and Clinical scales of the HCR-20 and VRS for a sample of 44 male inpatients. The ability of these risk assessments to predict recorded incidents of violent behaviour, as measured by official incident reports, was then assessed. The predictive validity of individual scale items was also considered. The HC composite and VRS did not predict inpatient violence within the first 6 months of admission. However, the clinical sub-scale of the HCR-20 was predictive of violence, abuse or harassment. When considering repetitiveness there was some indication across the scales that static factors predicted isolated incidents and dynamic factors repetitive violence. A number of individual items within the scales appeared to act as predictive or protective factors for inpatient violence. Despite the poor predictive validity of the scales overall, this study provides some indication of the differential utility of these structured clinical assessments for predicting short-term risk of violence in inpatients. In particular the use of dynamic clinical factors in identifying those likely to engage in imminent repetitive violence.