This study is to evaluate the sampling bias of a new cyclone (Tsai et al. 1999) and a 10 mm nylon cyclone under field sampling conditions. Both cyclones were operated at the optimum flow rate of 1.7 L/min. The Marple personal cascade impactor was used to measure the particle size distributions, which were used to calculate the reference RPM (Respirable Particulate Matter) concentrations according to the new ACGIH criteria. A wind tunnel was used in the field to ensure uniform particle concentration and wind speed across the tested cyclones and impactors in the working section. The wind speed, which was varied from 0.8 to 2.3 m/s, was kept constant in the wind tunnel. Near the packaging area of a glazing-making factory (mass median aerodynamic diameter, or MMAD: 11.1 μm), the average ratio of the RPM concentrations measured by the cyclones to the standard RPM concentrations was found to be 1.20 and 1.10 for the 10 mm nylon and the new cyclone, respectively. Near the packaging area of a lead powder factory (MMAD: 4.82 μm), the average ratio of the RPM measured by the cyclones to the reference RPM was found to be 1.13 and 1.02 for the 10 mm nylon and the new cyclone, respectively. Wind speed and total airborne particle concentration were found to have little effect on the accuracy of the two respirable cyclones. This field study showed that the new cyclone is more accurate than the 10 mm nylon cyclone based on the reference RPM.