<p>Limited information in China is available on how air pollution impacts daily hospitalizations for respiratory diseases before and under COVID-19 pandemic. Our aim is to investigate the potential effects of ambient air pollution and the pandemic on daily hospitalizations for respiratory diseases. Generalized additive model (GAM) are combined with Random Forest (RF) and Shapley Additive Interpretation (SHAP) method to analyze pollutant-hospitalization associations and rank pollutant importance based on a 9-year dataset from a tertiary grade A general hospital. Machine learning pinpointed PM_2.5, PM₁₀, and NO₂ as main drivers of increased hospitalizations, consistent with GAM risk estimates, while temperature as a covariate was second only to these three pollutants. During the whole periods, when PM_2.5 (8-day moving average, MA₀₇), PM₁₀ (MA₀₇), NO₂ (MA₀₇) and O₃-8&#xa0;h (MA₀₅) rose by 10&#xa0;μg/m³, daily hospitalizations increased by 2.19%, 1.35%, 2.84% and 0.94% respectively. It was 15.70% for every 1&#xa0;mg/m³ rise in CO (MA₀₇). Specifically, PM_2.5, PM₁₀, NO₂ and CO at MA₀₂ effects strengthened during the pandemic, with hospitalization increases rising from 1.11%, 0.65%, 1.82%, and 12.41% before pre-pandemic to 1.91%, 0.69%, 3.9%, and 22.07%, respectively. The associations varied by individual characteristics, with males susceptible to PM₁₀ and females to PM_2.5. The effects of PM_2.5, PM₁₀, and NO₂ in the warm season were more pronounced than in the cold, although they tend to have lower concentrations. This study indicates that exposure to PM_2.5, PM₁₀ and NO₂ has a stronger impact on respiratory health than to other pollutants.</p> Graphical Abstract <p></p>