Objective: The goals of this study were to establish a new model for predicting in-hospital mortality of percutaneous coronary intervention (PCI), and to compare the predictive accuracy of the new model with the European System for Cardiac Operative Risk Evaluation (EuroSCORE) model. Methods: We retrospectively analyzed 18206 Chinese patients undergoing PCI in our hospital between January 2002 and December 2008. Univariate and multivariate logistic regression analysis were performed to determine the preoperative risk factors that predicted in-hospital mortality. The EuroSCORE model was used to predict in-hospital mortality. To improve the predictive value of the EuroSCORE, we proposed a modified EuroSCORE that included some PCI-specific factors. Results: Multivariate logistic regression analysis indicated that left main trunk lesions, type C lesions, preoperative high-risk state, emergency PCI, and severe reduction of ejection fraction (<30%) were independent predictors of PCI mortality. Receiver operating characteristic (ROC) curve analysis showed that the area under the ROC curve for the modified EuroSCORE was larger than that for the EuroSCORE, however, there was no significant difference between the actual mortality rate and the mortality rate predicted by the two models. The modified EuroSCORE more accurately predicted the mortality rate in patients with a EuroSCORE score of 7-9 points, whereas it was no better than the standard EuroSCORE in patients with EuroSCORE scores of 1-3 or 4-6 points. Conclusion: The modified EuroSCORE improves acute risk assessment in PCI patients compared with the EuroSCORE, and the mortality rate predicted by the modified EuroSCORE in high-risk patients was close to the actual mortality rate.