Caltrans and other agencies rely on travel demand models (TDMs) to forecast outcomes of changes in the transportation system, such as highway expansions and long-range transportation plans. However, these TDMs are imperfect. They are not capable of analyzing certain kinds of projects (for example, bicycle and pedestrian facilities). In addition, agencies are increasingly interested in project outcomes like vehicle miles traveled (VMT), greenhouse gas emissions, safety, noise, and equity. TDMs are unable to provide forecasts for many of these outcomes. TDM forecasts may be inaccurate, notably for VMT effects of induced travel. The model may also be inaccurate due to inherent bias in its calibration or because of input and parameter decisions by agency staff. This project will investigate these issues by first delineating the capabilities of TDMs to accurately forecast different kinds of outcomes for different types of projects. This will lead to recommendations on where the use of TDMs is appropriate and where alternative tools may be used/developed. This project will also assess the usefulness of TDMs in regulatory settings by determining how susceptible TDM analyses are to bias introduced by the model operator. These findings will inform policymakers of the opportunities and limitations in relying on TDMs as regulatory tools.