The sampling efficiency of three types of personal cyclone has been experimentally measured at a sampling flow rate of 2.1 1 min-1. All samplers were based on the original Higgins and Dewell-type cyclone. Samplers were characterised using a polydisperse glass microsphere aerosol contained in a calm air chamber. A modified Aerodynamic Particle Sizer was used to determine the particle size distribution of the aerosol entering and leaving the cyclones. Two of the cyclone types were shown to be statistically identical within experimental errors, when characterised in terms of their d50 points. However, the third type showed a significant deviation in sampling characteristics from the others. It was possible to trace this to a manufacturing error, which has since been rectified. Comparison with previously measured cyclone efficiency curves using similar techniques was good, indicating that characterising aerosols samplers using the APS and polydisperse aerosols is at a stage where experimental errors are not dominated by the instrument, or the aerosol type.