Objective To construct and validate the risk prediction model of unplanned peripheral arterial catheter extubation in ICU patients.Methods Totally 300 ICU patients (the derivative group) with peripheral arterial catheter were selected and their clinical data were collected retrospectively. The risk factors of unplanned extubation were screened using univariate analysis and logistic regression analysis, and the risk prediction model was constructed.A total of 117 ICU patients with indwelling peripheral arterial catheters were selected to verify the model. Results Totally, 23.00% patients in the derivative group had unplanned extubation. Age≥65 years old, hypertension, venous thromboembolism, radial artery as the insertion site, and use of venous catheter for catheterization were risk factors for unplanned extubation (all P<0.05). The area under the ROC curve of the prediction model was 0.888. The validation results of the model showed that the area under the ROC curve was 0.903, with the Yoden index,sensitivity and specificity of 0.714, 86.2% and 85.2% respectively. Conclusion The risk prediction model constructed has good performance in identifying ICU patients at risk for unplanned peripheral arterial catheter extubation, which can provide evidence for medical staff to take preventive measures.