While several scientific studies have linked PM2.5 to decreased lung function, there is still some degree of uncertainty regarding which particulate physicochemical properties are most harmful. We followed a panel of 57 healthy schoolchildren (857 person-days) to investigate the associations between a wide variety of PM2.5 and lung function in Heshan, China in 2016 for three periods. We monitored the daily concentrations of mass, chemical composition, size, number, surface area, and volume of particulate mixture. Associations of lung function with various particle metrics were estimated using generalized estimating equations and unconstrained distributed lag models. Random forest model was used to compare the relative importance of exposure metrics. Immediate (lag 0) associations of PM2.5 and carbonaceous aerosols with reduced FEV1 and MMEF, and accumulation-mode particles with FEV1 were found. Slightly delayed (lag 1, 2) effects on PEF were particularly prominent for Aitken-mode particles. Possible cumulative (lags 0-2) effects of PM2.5 and carbonaceous aerosols on PEF and Aitken-mode particles on FEV1, MMEF, and PEF were observed. This study provides comprehensive evidence that the physicochemical properties of particulate mixtures are associated with reduced lung function in children. Organic carbon (OC) may be an important risk factor for the decreased lung function related to PM exposure.