Previous estimates of the magnitude of highway congestion costs have employed equations relating the external cost imposed on motorists by an additional vehicle to the speed of traffic flow or the ratio of traffic flow to the maximum capacity of the road. While those equations may be accurate for rural roads and expressways, they may be less accurate on city streets where delays at intersections are a dominant factor in congestion costs. This study replaces single speed-volume equations with a traffic simulation model that replicates the queuing of vehicles at traffic lights, the dispersion of platoons of vehicles as they move from one intersection to another, and the interaction of intersecting traffic flows on an urban street network. The model is used with actual Toronto road and traffic data to produce new estimates of congestion costs on specific streets during the morning rush hour. The model produces a surprisingly high average congestion cost during the morning rush hour and a poor correlation of the results with those that would be estimated for the same traffic flows by the single equation models. The simulation technique allows the calculation of congestion costs on a street-by-street basis, generating the detailed information that would be necessary for a complex congestion pricing scheme.