We study the role of networks for the matching of PhD graduates to universities. We build a novel dataset that allows us to track publication outcomes and co-author networks of both PhD graduates and their advisors. First, we find that the advisor’s pre-existing collaboration network has a substantial effect on which institution a PhD graduate matches to—even when comparing graduates within a fine-grained group defined by both background and hiring institution, and further controlling for topic overlap between the PhD dissertation and the potential hiring institution. Second, when comparing graduates from a similar background new hires with a network connection are more productive than hires without a connection. Network hires are systematically matched to more productive institutions, collaborate more with their new colleagues without changing research topics, and produce similar output as their new colleagues hired without a network connection at the same institution. Overall, the results document that the advisor’s collaboration network helps allocate graduates to research positions, with beneficial effects to the student and neutral effects to the hiring institution.