In-use, fuel-based motor vehicle emission factors were determined using measurements made in a highway tunnel in Pittsburgh, Pennsylvania. Concentrations Of PM2.5 mass, CO, CO2, and NO, were measured continuously. Filter-based measurements included PM2.5 mass, organic and elemental carbon (OC and EQ, inorganic ions and metals. Fuel-based emission factors for each pollutant were calculated using a fuel-carbon balance. The weekday traffic volume and fleet composition varied in a consistent diurnal pattern with the estimated fraction of fuel consumed by heavy-duty diesel ve ' hicle (HDDV) traffic ranging from 11% to 36%. The emission rate of most species showed a significant dependence on sample period. NOx, PM2.5, EC and OC emission factors were significantly larger during the early morning, truckdominated period. Emissions of particulate metals associated with brake wear (Cu, Sb, Ba and potentially Ga) were emitted at higher rates during the rush-hour period, which is characterized by slower, stop-and-go traffic. Emission rates of crustal elements (Fe, Ca, Mg, Li), Zn and Mn were highest during the early-morning period when there was more heavytruck traffic. A seasonal shift in average OC/EC ratio for the rush-hour period was observed; fall and summer OC/EC ratios are 1.0+0.6 and 0.26+0.06, respectively. Potential causes for this shift are increased partitioning of semi-volatile organic compounds into the gas phase during the summer months and/or effects of seasonal changes in fuel formulation. Emission factors for HDDV and light-duty vehicles (LDV) classes were estimated using a linear regression of emission factor as a function of fleet composition. The extrapolated emission factors generally agree with previously published measurements, though a substantial range in published values is noted. (c) 2006 Elsevier Ltd. All rights reserved.