Inertial filter, which classifies nanoparticles by inertial filtration, has been developed by Otani et al. (Aerosol and Air Quality Research, 7, 343-352 (2007) ) . The present work is aimed at formulating the fundamental design rules for the inertial filters with an appropriate cutoff size at a low pressure drop. We numerically calculated the classification performance for nanoparticles and compared the results with the experimental data obtained with TEM grids which has rigid and uniform structure as a model filter. As a result, thorough the theoretical calculation and experiments, (1) the collection efficiency of inertial filter can be successfully predicted by the conventional filtration theory based on the single fiber collection efficiency. (2) A filter with a finer fiber diameter can achieve smaller cutoff size at a lower pressure drop, however, the theoretical calculation suggested (3) an inertial filter consisting of fibers smaller than 14 μm cannot suppress the diffusional collection of nanoparticles with the diameter of 20 nm even at a filtration velocity of 30 m/s.