Abstract The satellite-based estimation of dry PM 2.5 mass concentration near the surface is a big challenge in the aerosol remote sensing fields, but urgently needed by the environmental monitoring. We present an experimental validation of a physical PM 2.5 remote sensing (PMRS) method which is not dependent on geographical location, based on ground-based remote sensing measurements at Jinhua City, a typical middle size city in East of China. The PMRS method is designed to employ currently available satellite remote sensing parameters as many as possible, including aerosol optical depth (AOD), fine mode fraction (FMF), planetary boundary layer height (PBLH) and atmospheric relative humidity (RH), and thus be capable of deriving PM 2.5 from instantaneous remote sensing measurements under different pollution levels. The key processes of the PM 2.5 method including size cutting, volume visualization, bottom isolation and particle drying are directly validated by comparing with reference parameters. We found that the size cutting of the PMRS method has a significant bias (about 0.86) resulting from the use of fine mode fraction to estimate PM 2.5 among all size of aerosol particles, which should be systematically corrected. The validation results of the volume visualization and particle drying of the PMRS method are quite satisfied while the bottom isolation procedure brings currently the maximum uncertainty to the PM 2.5 remote sensing. The improved PMRS method shows good performance on the remote sensing of hourly PM 2.5 with an average error of about 38 μg/m 3 and relative error of about 31%. The correlation coefficient between remote sensing and in situ hourly PM 2.5 is about 0.67 with a linear slope of 1.03 and intercept of 11 μg/m 3 while the means are very close (110.7 μg/m 3 versus 118.6 μg/m 3 ). The validation study also helps find out future improvement directions and demonstrates the possible application to ground-based remote sensing data.