Though rainfall is recognized as one of the main mechanisms to reduce atmospheric particulate pollution, few studies have quantified this effect, particularly the corresponding lag effect and threshold. This study aimed to investigate the association between rainfall and air quality using a distributed lag non-linear model. Daily data on ambient PM2.5 and PM2.5–10 (particulate matter with an aerodynamic diameter less than 2.5 μm and from 2.5 to 10 μm) and meteorological factors were collected in Guangzhou and Xi'an from 2013 to 2014. A better washout effect was found for PM2.5–10 than for PM2.5, and the rainfall thresholds for both particle fractions were 7 mm in Guangzhou and 1 mm in Xi'an. The decrease in PM2.5 levels following rain lasted for 3 and 6 days in Guangzhou and Xi'an, respectively. Rainfall had a better washout effect in Xi'an compared with that in Guangzhou. Findings from this study contribute to a better understanding of the washout effects of rainfall on particulate pollution, which may help to understand the category and sustainability of dust-haze and enforce anthropogenic control measures in time.