The problem of allocating the capacity of a road network is considered. The network is presented by a digraph with one source-sink pair and n alternative (not intersecting) routes. To arrive at the optimal strategy for a road network's capacity allocation, a bi-level optimization program is formulated. The upper level is a modeling decision made by the city administration entrusted with making constructive changes to the road network. The administration seeks to minimize the overall travel time on the network. The lower level is modeling the behavior of network's users, who react to any constructive changes. Each driver tends to minimize his/her own travel time from origin to destination. The optimal solution is obtained explicitly for the studied network. A methodological tool for decision-making support in road network sphere is then developed.