ISO dust injectors are generally applied to re-suspend standard dust particles, e.g. ISO A2, to challenge air filters used in internal combustion engines. The size distribution of the re-suspended dust particles is very important for acquiring the size fractional efficiency and loading performance of the filter. Generally, the existing size distributions provided by the powder manufacturer are directly applied or real-time aerosol instruments are used to measure the particles. Various instruments, however, report different size distributions for the same dust sample and also they are quite different with that provided,by the manufacturer. The main reasons are likely due to the particle transport loss during the sampling by the real-time instruments, especially for dust particles larger than 1 pm, as well as the different measurement principles and different sizing ranges of the instruments. In addition, the provided size distributions by the manufacturer are not ideal to represent the re-suspended dust particles because those distributions are measured in liquid when they are highly deagglomerated by the aids of chemical agents and ultra-sonication. In this study, therefore, the in-situ and noninvasive shadowgraph technique with an image analysis technique (particle/droplet image analysis, PDIA) were applied to measure the size of the re-suspended dusts at the moment they are being generated. The experimental system consisted of an 8 Mpixel CCD camera equipped with a high magnification micro-lens, up to 28 x, to allow the measurement of small particles down to 1.5 mu m. Monodisperse PSL particles with diameters of 5,17 to 26 mu m were generated from a home-built generator and used to validate the sizing accuracy of the system. The validated system was then applied to measure the size distribution of the widely used ISO A2 fine dusts re-suspended-by different dust dispersers, including the ISO light-duty and ISO heavy-duty injectors. Results showed that there was a large discrepancy between the size distributions determined by the powder manufacturer and those from ISO injectors by PDIA, indicating the ISO injectors did not deagglomerate A2 dusts well. This study first reported the reliable size distributions for A2 dusts re-suspended by ISO injectors under different operation conditions. (C) 2017 Elsevier B.V. All rights reserved.