BACKGROUND: Smaller sizes of ambient particulate matter (PM) can be more toxic and can be breathed into lower lobes of a lung. Children are particularly vulnerable to PM air pollution because of their adverse effects on both lung functions and lung development. However, it remains unknown whether a smaller PM has a greater short-term impact on childhood pneumonia. AIMS: We compared the short-term effects on childhood pneumonia from PM with aerodynamic diameters ≤1 μm (PM(1)), ≤2.5 μm (PM(2.5)), and ≤10 μm (PM(10)), respectively. METHODS: Daily time-series data (2016-2018) on pneumonia hospitalizations in children aged 0-17 years, records of air pollution (PM(1), PM(2.5), PM(10), and gaseous pollutants), and weather conditions were obtained for Hefei, China. Effects of different PM were quantified using a quasi-Poisson generalized additive model after controlling for day of the week, holiday, seasonality and long-term time trend, and weather variables. Stratified analyses (gender, age, and season) were also performed. RESULTS: For each 10 μg/m(3) increase in PM(1), PM(2.5), and PM(10) concentrations over the past three days (lag 0-2), the risk of pneumonia hospitalizations increased by 10.28% (95%CI: 5.88%-14.87%), 1.21% (95%CI: 0.34%-2.09%), and 1.10% (95%CI: 0.44%-1.76%), respectively. Additionally, both boys and girls were at risk of PM(1) effects, while PM(2.5) and PM(10) effects were only seen in boys. Children aged ≤12 months and 1-4 years were affected by PM(1), but PM(2.5) and PM(10) were only associated with children aged 1-4 years. Furthermore, PM(1) effects were greater in autumn and winter, while greater PM(2.5) and PM(10) effects were evident only in autumn. CONCLUSION: This study suggests a greater short-term impact on childhood pneumonia from PM(1) in comparison to PM(2.5) and PM(10). Given the serious PM pollution in China and other rapid developing countries due to various combustions and emissions, more investigations are needed to determine the impact of different PM on childhood respiratory health.