Objective To investigate the predictive value of the prognostic nutritional index (PNI) for 28-day all-cause mortality in patients with chronic obstructive pulmonary disease (COPD) in intensive care unit (ICU). Methods The relationship between PNI and short-term mortality in COPD patients was analysed using COX proportional hazards and restricted cubic spline (RCS) models. Receiver operating characteristic (ROC) curves were plotted and area under the ROC curve (AUC) was calculated to assess the predictive performance of PNI. The optimal cut-off value for PNI was determined using the Youden index, and the data were divided into a low PNI group and a high PNI group. Kaplan-Meier curves were then constructed and the log-rank test was used to assess differences in survival between the two groups. Results A total of 980 COPD patients were included in the study. Multivariable COX regression analysis showed that PNI was an independent factor influencing short-term mortality in the severe COPD patients (HR=0.972, 95%CI 0.948 - 0.995, P=0.019). RCS curve results showed a non-linear relationship between PNI and short-term mortality in the severe COPD patients (P for non-linear=0.032), with the risk of death gradually decreasing as PNI increased. The ROC curve indicated that PNI had some predictive power, comparable to that of SOFA score [(AUCPNI=0.693) vs. (AUCSOFA=0.672)]. Kaplan-Meier curve analysis showed a significant difference in survival time between the low (≤38.3) PNI group and the high (>38.3) PNI group (P<0.05). Conclusions PNI has a certain predictive role for short-term all-cause mortality in patients with severe COPD. Patients with low PNI at ICU admission have a higher risk of short-term mortality.