This paper estimates for 28 product groups a characteristic parameter that reflects the topological structure of its trading network. Using these estimates, it then describes how the structure of international trade has evolved during the 1980-2000 period. Thereafter, it demonstrates the importance of networks in international trade by explicitly accounting for their scaling properties when testing the prediction of the Heckscher-Ohlin model that factor endowment differentials determine bilateral trade flows. The results suggest that differences in factor endowments increase bilateral trade in goods that are traded in "dispersed" networks. For goods that are traded in "concentrated" networks, factor endowment differentials are less important.