BACKGROUND: The diagnosis of chronic obstructive pulmonary disease (COPD) typically relies on spirometric measurements. The Staging of Airflow Obstruction by Ratio (STAR) classification, a newly proposed system for grading the severity of pulmonary function, has been suggested as a potentially better predictor of outcomes than other classifications. However, the long-term stability of the STAR classification, especially in comparison to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) classification, remains unclear. METHODS: In this retrospective cohort study, we analysed data from 622 patients with COPD, enrolled in a pay-for-performance programme at two university hospitals in Taiwan. Patients were classified according to the GOLD and STAR classifications, based on post-bronchodilator spirometry results. The study assessed the agreement between these classifications and the stability of each over a 3-year period, categorising patients into four patterns: no change (stable stage throughout), progression (persistent shift to severe stage), instability (fluctuating between higher and lower stages) and reversal (sustained improvement to a less severe stage). RESULTS: The STAR classification system identified a higher proportion of patients with instability or reversal patterns (42.1%) compared with the GOLD classification (31.0%). While fair coherence was noted between the two classifications over 3 years, the STAR classification demonstrated greater variability. Compared with the GOLD classification, the STAR classification exhibited a higher proportion of instability or reversal patterns in stage 2 but a lower proportion of these patterns in stage 4. CONCLUSION: Compared with the GOLD classification, the STAR classification demonstrated higher instability and reversal patterns, suggesting the need for careful consideration for its use in long-term COPD management. Further research is required to explore the clinical implications of these findings and to refine the use of these classifications.