The current prospective risk assessment study evaluated the application of the Chinese translation of the Historical-Clinical-Risk Management-20 Version 3 (HCR-20 V3 ) in a sample of 152 offenders with mental disorders and civil psychiatric patients. The ratings of the presence and relevance of risk factors were compared, as well as summary risk ratings (SRRs), both across offenders and civil psychiatric patients, and across male and female sub-samples. Interrater reliability was consistently "excellent" for the presence and relevance of risk factors and for SRRs. Concurrent validity analyses indicated that HCR-20 V3 was strongly correlated with Violence Risk Scale (from r = 0.53 to 0.71). The results of predictive validity analyses provided strong support for the bivariate associations between the main indices of HCR-20 V3 and violence within 6 weeks, 7-24 weeks, and 6 months; SRRs added incrementally to both relevance and presence ratings across three follow-up lengths.