Abstract: In order to investigate the impact of air pollution episode characterized by persistent high PM_2.5 exposure on people's hospitalization in Beijing, we collected data of air pollutants, meteorological factors and hospital admissions in Beijing from 2013 to 2018. Under six persistent PM_2.5-polluted scenarios defined by PM_2.5 concentration (75.0 μg/m³, 150.0 μg/m³, P₉₅ (205.8 μg/m³)) and duration (≥2 d, ≥3 d), generalized additive models (GAM) based on quasi-Poisson regression were used to carry out time series analyses to obtain the impact of air pollution episode characterized by persistent high exposure to PM_2.5 on hospitalizations of the entire population and subgroups. The results showed that: (1) For the entire population, relative to non-specified pollution scenarios, when the PM_2.5 exposure concentration was >150 μg/m³ and lasted for two days or more, the hospitalization risk for non-accidental diseases and cardio-cerebrovascular diseases increased by 5.0% (95%CI, 1.2% to 9.0%) and 5.6% (95%CI, 1.8% to 9.5%), respectively. (2) A stratified analysis of subgroup populations found that under heavy PM_2.5-pollution scenarios that lasted for two days or more, non-accidental and cardio-cerebrovascular hospitalizations significantly increased in the subgroups of men, women, people 0-64 years old, people 65-74 years old, and people 75 years or older. In the extreme PM_2.5-pollution concentration (>205.8 μg/m³) that lasted for three days or more, non-accidental and cardio-cerebrovascular hospitalizations significantly increased in only the subgroups of women, people 0-64 years old and people 65-74 years old. (3) For respiratory admissions, we only detected statistical significance when PM_2.5 exposure concentration was greater than 150.0 μg/m³ and lasted for two days or more in the 0-64 years old population, and the risk increased by 3.4% (95%CI, 0.2% to 6.6%). This study shows that air pollution episode characterized by persistent high exposure to PM_2.5 had a significant impact on hospital admissions.