Light-duty vehicle emissions were measured at the Caldecott Tunnel in August and October 1994. In the interval between these two periods, the average oxygen content of gasoline sold in the San Francisco Bay area increased from 0.3 to 2.0% by weight. Compared to the August (low-oxygenate) sampling period, measured pollutant emission rates during the October (high-oxygenate) sampling period for CO and VOC decreased by 21 ± 7 and 18 ± 10%, respectively, while NOx emissions showed no significant change. Formaldehyde emissions increased by 13 ± 6%, acetaldehyde emissions did not change significantly, and benzene emissions decreased by 25 ± 17%. Speciated VOC emission profiles show that the use of oxygenated gasoline resulted in higher MTBE and lower aromatic hydrocarbon emissions, higher isobutene, and lower aromatic aldehydes. The normalized reactivity of NMOG emissions did not change significantly between the low-oxygenate and high-oxygenate sampling periods. VOC exhaust speciation profiles for vehicles operating in the hot-stabilized mode at the Caldecott Tunnel match the speciation profile for cold-start emissions from new vehicles as measured in the Auto/Oil program. California's motor vehicle emission factor model, EMFAC7F, accurately predicts the VOC/NOx ratio measured at the Caldecott Tunnel in August, but underpredicts the observed CO/NOx ratio by a factor of 1.5−2.2 over the range of vehicle speeds observed at the tunnel.