Vehicle classification information is crucial input to transportation facility design, operations, and planning. As freight transportation is becoming more and more critical to regional and national economies, freight modeling is now an imperative issue for many transportation management agencies, for which truck classes and volumes are key input. The current state-of-the-practice vehicle classification methods rely on fixed location sensors such as pneumatic tubes, loop detectors (or other types of magnetic sensors), video cameras, besides manual observation and classification. These existing classification methods are either too expensive to be deployed for large areas or subject to errors under specific situations. It is currently a challenging issue to automatically and accurately classify vehicles in a low-cost manner for large areas including both freeways and arterials. This research examines the feasibility of using mobile traffic sensors for automatic vehicle classification. Mobile sensors include GPS and other types of tracking devices that can trace the movement of individual vehicles. They can provide information (e.g. vehicle traces) that promises great advances in many science and engineering fields, including public health monitoring/diagnostics, extraction of personal or social behaviors, and transportation. The proposed research focuses on developing models and algorithms that can classify vehicles based on short vehicle traces collected from mobile sensors. It represents the first step towards this direction by investigating the feasibility of using mobile data for vehicle classification. The proposed method is to characterize the features of the short traces of different vehicle types, based on which to develop learning algorithms to distinguish vehicle classes using the features. The proposed research is of significant importance to the region. The New York State and especially the New York City (NYC) has significant freight traffic flow. Statewide, there are a few freight corridors such as the I-95 corridor that are crucial to the economy and everyday life of the region and our entire nation. Integrating freight movement in regional demand modeling or even developing specific freight demand models is currently considered or being developed by transportation management agencies in the region. The State now relies on pneumatic tubes and WIM stations to collect vehicle classification data. Alternative techniques for accurate vehicle classification information, as the proposed research aims to address, will help improve the current vehicle classification practice in the region and help the development and calibration of regional freight demand models. This will ultimately facilitate policy makers to make informed decisions on operating/managing the region’s transportation system. The proposed study also contributes to the core of the UTRC2’s Theme on “Planning and Managing Regional Transportation Systems in a Changing World.” Vehicle classification data are crucial to the planning and management of urban transportation systems. Emerging techniques such as mobile sensors provide new directions to solve current challenges; the proposed research seeks new perspectives and alternative solutions to traditional problems.