Objective To explore the predictive value of the Geriatric Nutritional Risk Index (GNRI) for 30-day readmission in patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD).Methods The clinical data of 324 AECOPD patients were retrospectively collected. According to whether they were readmitted within 30 days of discharge, they were divided into 269 cases in the non-readmission group and 55 cases in the readmission group, and the clinical data of the two groups were compared.The influencing factors of 30-day readmission in AECOPD patients were analyzed by using logistic regression analysis.ROC curves were used to analyze the assessed value of 30-day readmission by GNRI alone as well as after calibrating the confounders.The practical value of predicting 30-day readmission was analyzed by Decision Curve Analysis (DCA) by GNRI alone as well as after calibrating the confounders.Results The percentage of long-term oxygen therapy, CAT score, WBC, and erythrocyte sedimentation rate were significantly higher in the readmission group than those in the non-readmission group, and its FEV1%pred, PaO2, and GNRI were significantly lower than those in the non-readmission group (all P＜0.05).Quartiles of GNRI divided the patients into 4 groups, the readmission rates in GNRI Q4, Q3, Q2, and Q1 group were 4.94% (4/81), 6.17%(5/81), 23.46%(19/81) and 33.33%(27/81), and the difference was statistically significant (P＜0.05).Logistic regression analysis showed that, long-term oxygen therapy, FEV1%pred, CAT score, and GNRI were the influencing factors for 30-day readmission in AECOPD patients (all P＜0.05).Before and after calibrating for confounders of age, gender, long-term oxygen therapy, CAT score, and FEV1%pred, the risk of readmission was significantly lower in those with GNRI Q3 and Q4 (all P＜0.05), and the OR value showed a decreasing trend (P trend＜0.05).The AUC of GNRI alone predicting 30-day readmission in AECOPD patients was 0.819 (95%CI:0.773-0.860), and the AUC after calibrating for confounders was 0.858 (95%CI:0.815-0.894), and the diffe-rence between the AUCs before and after calibrating was not statistically significant (P＞0.05).The net benefit of GNRI alone in predicting 30-day readmission was 0.04-0.61, which was 0.05-0.70 after calibrating for confounders.Conclusion GNRI is an independent influencing factor of 30-day readmission in AECOPD patients, and it can effectively predict their occurrence of 30-day readmission.