Road users have heterogeneous preferences regarding travel choices, such as time-of-day to travel, routes, and willingness to carpool. Understanding the variation in preferences would allow better modeling of the temporal distribution of travel demand and the development of pricing policies. This study examines the demographic and socioeconomic profiles of motorists at different times of day at six locations in southern California by using on-road remote-sensing measurements and license plates images obtained in 2007 and 2008 by the California South Coast Air Quality Management District (SCAQMD). The remote sensing data provide license plate images in addition to speed, time to the second, and emissions. By matching the license plate images to California Department of Motor Vehicles (DMV) registration records, anonymized individual vehicle ownership records can be obtained. The missing demographic and socioeconomic profiles of vehicle owners are supplemented with census data at the spatially fine level of census block group. In contrast to person-based methods like traditional travel surveys, remote sensing methods do not track individuals’ behavior over a day (or longer), but provide several advantages: 1) the precise time log of the trips; 2) fewer missing data since correct license numbers are registered for most of the vehicles passing a monitoring site; and 3) data acquired for any desired length of monitoring time at minimal cost. From 98 remote-sensing emissions monitoring sites, located for the SCAQMD at highway entrances and interchanges with the hope of identifying gross emitters, six sites were chosen for this study for a detailed analysis. Three of these six sites were in Los Angeles County and three in Orange County. The selected sites experienced some of the worst congestion in Los Angeles and Orange counties. The neighborhoods around the study sites represent different land use characteristics, such as residential, shopping, commercial and industrial. By examining the speed patterns of the highway ramps at the monitoring sites, peak hours at four of the six study locations have been identified. Not all the study locations share the same peak-hours pattern, however. Among the four locations with varying speed patterns over a day, three of them were compared with the highway mainline speed data in close proximity to the study sites. The two sets of speed data show close correlations. The peak hours used in this study were based on the actual speed patterns at ramps. This study focuses on the differences of the profiles of motorists during peak and non-peak hours. This study uses a continuous-time survival model to analyze the varying demographic and socioeconomic profiles of motorists in the weekdays in Los Angeles and Orange counties. The availability of the precise time of passing vehicles makes it appropriate for using a continuous time approach. The extended Cox model with a non-parametric baseline hazard is used to accommodate both the time-invariant and time-varying effects of the covariates. The empirical results indicate that those who travel in the peak hours are more likely to be younger, to have lower education attainment, to have lower median household income, more likely to be non-English speakers and to be non-white, and less likely to be in management and professional occupations. This study demonstrates an application of the remote sensing data to the analysis of the costs of traffic congestion at a spatially and temporally disaggregated level. The results show that the burden of congestion is not evenly distributed among income groups. The congestion cost borne by the bottom income group is comparatively small even though they represent the highest share of the total observations at some study locations.