The lack of standardized violence risk assessment instrument has disadvantaged nurses in clinical practice in China, where violent behaviour is an increasing problem. This study conducted a validation of the Brøset Violence Checklist that has proven effective in violence risk prediction in other countries. A sample of 296 patients consecutively admitted to two wards of a psychiatric hospital in Beijing was recruited. These patients were assessed on day shift and evening shift for the first seven days of hospitalization. Violence data and preventive measures were concurrently collected from nursing records and case reports. A total of 3707 assessments for 281 patients were collected revealing 93 episodes of violence among 55 patients. Receiver operating characteristics yielded an area under the curve of 0.85. At the cut-off point of one, its sensitivity/specificity was 78.5%/88.2% and the corresponding positive/negative predictive value was 14.6%/99.4%. In some false positive cases, intense preventive measures had been implemented. Positive feedback from the nurses was gained. The Brøset Violence Checklist was proved as an easy-to-use and time-saving instrument, therefore, regarded as a promising tool to determine if the psychiatric users are potentially violent in the short term.