Objective To explore the optimal high risk cut-off point of Caprini Risk Assessment Scale in non-surgical patients admitted to ICU.Methods A retrospective case-control study was conducted to recruit 154 non-surgical ICU patients who had a confirmed deep venous thrombosis (DVT) diagnosis as the case group, and another concurrent 154 non-surgical, non-DVT patients as the control group.The receiver operating characteristic curve (ROC) was run to identify the accuracy and optimal high risk cut-off point of the Caprini Risk Assessment Scale.Results The area under ROC curve was 0.709 and 95% confidence interval was 0.655 to -0.759. With a cut-off point of 7,Youden′s index was the largest,at 0.331; the sensitivity was 70.78%, the specificity 62.30%,positive predictive value （PPV） was 65.27%,negative predictive value （NPV） was 68.09%.Conclusion A cut-off point of 7 for the Caprini Risk Assessment Scale is optimal for non-surgical ICU patients to screen high-risk cases.