<p>Studies report a high 28-day all-cause mortality rate in patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD). The specific prognostic value of lymphocyte subsets for critically ill patients remains unclear. We aimed to evaluate the association between lymphocyte subsets and 28-day all-cause mortality in patients with AECOPD admitted to the intensive care unit (ICU). We conducted a retrospective cohort study on patients with severe AECOPD admitted to our hospital from January 2022 to December 2025. Comparisons were made between the survivors and non-survivors. We used restricted cubic splines, and Receiver Operating Characteristic (ROC) curves to describe the association between lymphocyte-related indicators and 28-day all-cause mortality. To prevent overfitting and robustly identify independent risk factors, Boruta feature selection and LASSO regression were sequentially utilized, followed by multivariable logistic regression to construct a prognostic model. The study enrolled 382 patients. During the 28-day follow-up, 115 (30.1%) patients died, and 267 (69.9%) survived. Among all examined immune cells, the CD3 + CD4+ T lymphocyte count demonstrated the strongest predictive performance. CD3 + CD4+ T lymphocytes exhibited the best predictive performance with a cutoff value of 157.61/µL. A prognostic model incorporating CD3 + CD4+ T lymphocytes and other key clinical features showed excellent discriminative ability with an area under the curve (AUC) of 0.860. A marked decline in CD3 + CD4+ T lymphocytes (&lt; 157.61/µL) is a significant and readily accessible independent risk factor for 28-day all-cause mortality in patients with AECOPD admitted to the ICU. A nomogram based on the multivariable logistic regression model provides a practical tool for early risk stratification and targeted immune monitoring in critical care settings.</p>