A new approach and experimental technique are proposed to determine times of metal particle combustion in flames of polydisperse aerosols. Laminar flames are produced in air at 1 atm, using aerosol jets formed by an electrostatic particulate method. The flame radiation intensities as a function of vertical coordinate are measured and compared with the flame radiation profiles reconstructed using experimental data and simplified models. The experimental data used include particle size distributions, flame velocities, and temperatures of metal ignition and combustion. The simplified models describe the particle ignition delay, combustion time, and particle flame radiation intensity as a function of particle diameter, D. Variable parameters of the models describing particle radiation intensities and combustion times are adjusted to achieve the best fit between the reconstructed and measured flame radiation profiles. A set of parameters providing the best agreement between the reconstructed and measured profiles is selected for several aerosol flames produced by powders of different sizes of the same material. These parameters are assumed to adequately describe particle combustion times and radiation intensities for the chosen material. The experimental radiation profiles for both aluminum and magnesium aerosol flames with particles of different sizes were found to be in very good agreement with the respective reconstructed profiles. For both metals, particle radiation intensities were well described by a D-3-type expression. The combustion times for magnesium aerosol particles were well described by the traditional D-2-law with the evaporation constant close to those reported earlier for single particles. Aluminum aerosol particle combustion was better described by a D-1-law and combustion times of fine (<80 mum) aluminum particles in the aerosol were somewhat longer than the reported earlier combustion times for single aluminum particles. (C) 2003 The Combustion Institute. All rights reserved.