Carbapenem-resistant organisms (CROs) are associated with considerable mortality clinically. There is a lack of effective tool to predict individual prognosis. We aim to determine if host immunity can be utilized to predict the prognosis of patients infected with CRO. From December 2018 to August 2019, we recruited CRO-infected patients to evaluate risk factors for 30-day mortality. Clinical, routine laboratory, immune and microbiological features were investigated and subjected to univariate and multivariate analyses. The final predictive models were established based on the regression coefficients of multivariate logistic regression. A total of 127 CRO-infected patients were enrolled in our study, including 85 survivors and 42 non-survivors. The number and IFN-gamma producing ability of lymphocytes were remarkably decreased in non-survivors. The number of IFN-gamma(+)CD4(+)T cells could effectively predict 30-day mortality of CRO infection. Its area under the receiver operating characteristic (ROC) curve, sensitivity, specificity and accuracy, were 0.889 (95% confidence interval [CI], 0.834-0.945), 81.0, 80.0, and 80.3%, respectively. In multivariate analysis of laboratory parameters, IFN-gamma(+)CD4(+)T cell number and creatinine concentration were selected for the 2-marker model to predict prognosis fleetly. Its area under the ROC curve, sensitivity, specificity and accuracy were 0.894 (95% CI, 0.841-0.947), 83.3, 82.4, and 82.7%, respectively. Impaired lymphocyte function was an important factor to affect the outcome of CRO-infected patients. A 2-marker model based on the combination of IFN-gamma(+)CD4(+)T cell number and creatinine showed good performance in predicting the prognosis of CRO infection.