This dissertation furthers the understanding of traffic dynamics in urban street networks. It focuses on networks that exhibit traffic inhomogeneities, to understand if parts or all of these networks can be described and controlled through aggregate relations. Using an idealized grid network, the work shows that spatially-inhomogeneous traffic is more commonplace than previously thought. Even when Origin-Destination (OD) demand patterns are spatially homogeneous, higher flows are observed in the center of the network, and lower flows in outlying portions. This spatial pattern persists when the network becomes congested. As congestion propagates, only traffic in the network’s center area remains homogeneous. Street links in the outlying portions exhibit levels of congestion that vary by travel direction, and therefore cannot be described using aggregate relations. Despite the presence of traffic inhomogeneities, it is shown that cordon control improves network performance when used to regulate vehicular accumulation in the network’s center. However, the effectiveness of this control method is greatly determined by features of the O-D table. Multiple fixed cordon sizes are tested. Larger cordons, farther away from the center, are shown to produce greater improvements. The effectiveness of cordon metering is compared against alternative means of congestion control and in more realistic settings to confirm present findings.