In this study, two micro-orifice-based cascade impactors, including the micro-orifice uniform deposit impactor (MOUDI, MSP Model 110) and the NCTU micro-orifice cascade impactor (NMCI), were tested for the collection efficiency and interstage loss of nanoparticles. In the NMCI, new nozzle plates with smooth nozzle shape made by the LIGA (Lithography, Electroplating, and Molding) process were used to replace the 7th-10th stages in one of the MOUDI. Test results show that after adjusting proper S/W ratios (S: jet to plate distance, W: nozzle diameter) to 2.52, 3.01, 13.44, and 24.75 for the 7th, 8th, 9th and 10th stage of the NMCI, respectively, and 5.56, 11.18, 9.3, and 10.9 for the 7th, 8th, 9th and 10th stage of the MOUDI, respectively, the cutoff aerodynamic diameters (dpa50) are close to the nominal values given in Marple et al. (1991). Different S/W ratios are needed due to differences in the nozzle shape and nozzle diameter between two cascade impactors. Total interstage loss of nanoparticles from the inlet to the 6th-10th stage of the MOUDI exists due to the convection-diffusion mechanism, which increases with decreasing dpa. For the MOUDI, total loss is 2.9-15.3 % (dpa: 105.8 to 15.4 nm) for the inlet to the 6th stage and it increases to 20.1-26.1 % (dpa: 23 to 15.4 nm) for the inlet to the 10th stage, respectively. Similar but slightly lower loss also exists in the NMCI. Field comparison tests in the ambient air show that mass size distributions measured by the MOUDI agree well with those of the NMCI. Finally, nozzle clogging tests using high concentration incense smokes indicate that the NMCI has a much less tendency for particles to clog in the nozzles than the MODUI.