High levels of clustering—the tendency for two nodes in a network to share a neighbor—are ubiquitous in economic and social networks across different applications. In addition, many real-world networks show high payoffs for nodes that connect otherwise separate network regions, representing rewards for filling “structural holes” in the sense of Burt (1992) and keeping distances in networks short. This paper proposes a parsimonious model of network formation with introductions and intermediation rents that can explain both these features. Introductions make it cheaper to create connections that share a common node. They are subject to a tradeoff between gains from shorter connections with lower search cost and losses from lower intermediation rents for the central node. Stable networks are shown to have high levels of clustering at the same time that they permit substantial intermediation rents for nodes bridging structural holes.