Atmospheric particulate pollution is an important risk factors for stroke. The aim of this study was to investigate the association of size-specific particulate matter (PM) and their components with stroke hospitalization in Northwest China's arid area from 2018 to 2022. Stroke hospitalization data were collected from Gansu Province. Size-specific PM with chemical speciation profiles and concurrent meteorological Data spatially aligned with the geographical coordinates of the included hospitals. The space-time-stratified case-crossover design and the conditional quasi-Poisson regression model were used to analyze the association between PM and stroke hospitalization, and stratified analysis was conducted based on age, gender, region, season and stroke type. This study included 759,473 stroke patients. PM1, PM1-2.5, PM2.5 and PM10 were all observed the highest risk of stroke hospitalization at lag 06 day, with RRs of 1.238 (95 % CI: 1.217-1.260), 1.057 (95 % CI: 1.049-1.066), 1.067 (95 % CI: 1.060-1.074) and 1.012 (95 % CI: 1.009-1.015). Among PM components, black carbon (BC) demonstrated the most pronounced association with stroke hospitalization risk (PM2.5, RR = 1.045 per 1 mu g/m3 increase, 95 % CI:1.037-1.054; PM1 and PM10, RR = 1.580 per 1 mu g/m3 increase, 95 % CI:1.535-1.626). Subgroup analyses further indicated greater vulnerability among women, the elderly, during the cold season, and for ischemic stroke (IS) compared to hemorrhagic stroke (HS). Our findings indicate strict air quality intervention measures based on size-specific effect of PM should be taken to reduce the burden of stroke.