Objective To construct and validate the risk prediction model of infection in patients with liver cancer after transcatheter arterial chioibolization (TACE). Methods Clinical data of 1043 liver cancer patients underwent TACE in a hospital from January 2016 to March 2018 were collected retrospectively. According to whether infection occured after TACE, patients were divided into infected group and non-infected group. Risk factors for the infection after TACE were screened by logistic regression analysis, and nomograph model was constructed and internally verified by Bootstrap method. 349 liver cancer patients underwent TACE from January to July 2020 in this hospital were collected as the validation set for external validation of the model. C-statistics and calibration curves were used to evaluate the prediction performance of the model. Results Among 1043 liver cancer patients, 105 were infected after TACE, thus infection incidence was 10.07%. History of biliary tract interventional surgery, largest tumor diameter >5 cm, ascites, tumor rupture hiorrhage, dosage of lipiodol >12 mL and the use of additional ibolic materials were independent risk factors for the infection after TACE (all P < 0.05). Previous TACE history was the protective factor for the infection after TACE (P < 0.05).The risk prediction model for the infection after TACE in liver cancer patients were Logit(P)=-4.403+1.874×(history of biliary tract interventional surgery) -1.101×(history of previous TACE surgery) +1.253×(largest tumor diameter >5 cm) +1.163×(ascites) +1.346×(tumor rupture hiorrhage) +0.901×(dosage of lipiodol >12 mL) +1.369×(use of additional ibolic materials). C-statistic of the model was 0.857 (95% confidence interval [CI]: 0.822-0.892). The maximal index of Youden, critical value, sensitivity and specificity were 0.596, 0.109, 84.8%, and 74.8% respectively. C-statistics of internal and external validation were 0.847 and 0.829 respectively. Calibration curve and Brier score showed that the model fit well. Conclusion The prediction model of infection risk of liver cancer patients after TACE constructed in this study has good discrimination and calibration, thus can be used to predict the infection of liver cancer patients after TACE.