The primary objective was to characterize exhaust aerosols from a small group of in-use, light-duty, spark ignition (SI) vehicles operated on-road, and on a chassis dynamometer. A significant particle signature above background was not measured at highway cruise condition. Number emissions were much higher during acceleration, at high-speed cruise, and during cold-cold starts. Group average fuel-specific number emissions range from 3.9 x 10(14) to 1.0 X 10(17) particles/kg of fuel. Cold-cold start temperatures, driving cycles and vehicular condition influence SI emissions. Elemental carbon was a major contributor to mass emissions measured in chassis dynamometer Unified Driving Cycle (UDC) tests averaging 64% and 34% of the mass emissions for cold-cold and hot start cycles, respectively. Average ratios of cold-cold to hot start emissions were 3.3, 7.6, and 22 for CPC number, filter mass and SMPS volume, respectively. Apportionment results showed that on a weekly weighted basis and on weekdays, the majority of observed particle number was attributed to heavy-duty diesel traffic. Weekend production of particles was attributable to light-duty automobiles. On a per vehicle basis, heavy-duty vehicles produced substantially greater number concentrations. On a fuel-specific basis, heavy-duty vehicles produce slightly higher concentrations of particles than light-duty vehicles. The relative contribution of light-duty vehicles to particle number emissions increased as particle size decreased, for the smallest particles apportioned number emissions were 1.3 x 10(16) and 7.1 x 10(15) particles/kg of fuel for heavy-duty and light-duty vehicles, respectively. Comparison of on-road chase and apportionment results with chassis dynamometer tests in a certification type facility suggests that the latter may underestimate real-world number emissions. (C) 2005 Published by Elsevier Ltd.