This paper provides experimental evidence on the economic determinants of intermediation networks by considering two pricing rules and three group sizes of subjects – 10, 50 and 100 subjects. When brokerage benefits accrue only to traders who lie on all paths of intermediation, stable networks involve interconnected cycles, and trading path lengths grow while linking and payoff inequality remain modest as the number of traders grows. When brokerage benefits are equally distributed among the traders located on the shortest paths, stable networks contain a few hubs that provide the vast majority of links, and trading path lengths remain unchanged while linking and payoff inequality explode as the number of traders grows.