27), possibly due to the low precipitation rates, had the highest monthly totals. TRA showed that local sources (within a 500 km radius) were the ones contributing the most to PM concentration in the period studied, totaling 55%, which allows us to point to vehicle and industrial emissions near the city of Limeira as the main sources. Keywords Trajectories • HYSPLIT model • Air mass residence time • Air pollution 1 IntroductionAir pollution occurs when a substance present in the atmosphere, resulting from natural (Morris & Therivel, 2009) or anthropogenic activities, can cause detrimental effects on humans, animals, vegetation, and even materials (Kampa & Castanas, 2008;Lelieveld et al., 2015;Feng et al., 2019). Anthropogenic sources, such as fossil fuels, industrial processes, and construction, are the main causes of air pollution (Drumm et al., 2014;Seinfeld & Pandis, 2006). According to Fajersztajn et al. (2017), air pollution is among the main health risks worldwide, and, in 2050, deaths related to air pollution may overcome those due to lack of sanitation and malaria. Consequently, exposure to air pollutants has severe health effects, such as mortality Abstract Emitted from vehicles, plant biomass combustion, and industries, particulate matter (PM) is an air pollutant widely studied by the scientific community due to its health effects (cardiorespiratory diseases, cancers, eye irritations, among others). The present study evaluates periods with high PM concentrations, defined as high-PM 10 episodes (daily concentrations above the 75th percentile), to define and assess the main possible sources of PM emission in the city of Limeira, São Paulo, Brazil. To determine the location of such sources, the trajectory regression analysis (TRA) statistical tool was used, based on trajectories obtained from the HYSPLIT model. The 75th percentile was calculated at 41.21 µg/m 3 , with a maximum concentration of 114.38 µg/m 3 . Results point to autumn, winter, and spring as the seasons with the highest number of episodes, accounting for