This paper incorporates owners' decisions to keep, repair or scrap their old vehicles into a simulation model of fleet emissions. This decision depends critically on the owner's perceived value of the vehicle, so we examine the factors affecting owners' valuations of their old vehicles using a unique longitudinal dataset. Willingness to accept for the vehicle is well predicted by mileage and condition of the car, and declines systematically with its age. Our estimated model of vehicle value is used as an input into a simulation model of a 1,000-car fleet representative of California's fleet. Other inputs into the simulation models are the estimated distributions of emissions in the fleet, and two equations that link emissions reductions to the cost of repairs. The simulation model is used to examine the role of scrap policies alone and combined with other policies for reducing emissions, such as current I/M programs and proposed emissions fees, and the welfare implications of combining such programs. The model incorporates both technical and behavioral relationships, and assumes that of all possible options (repairing the car, scrapping the vehicle, or paying the emissions fee without repairing the vehicle) the owner chooses the one with the least cost. We find that old car scrap programs may increase net welfare under a regulatory program like I/M in practice today, but that a stand alone scrap program is unlikely to provide very much in the way of emission reductions.