Using data on U.S. universities, we show that universities that give higher royalty shares to faculty scientists generate greater license income, controlling for other factors including university size, quality, research funding, and local demand conditions. We use pre-sample data on university patenting to control for the endogeneity of royalty shares. The incentive effects are larger in private universities than in public ones, and we provide survey evidence on performance-based pay, government constraints and objectives of Technology License Offices that helps explain this finding. Royalty incentives work through two channels -- raising faculty effort and sorting scientists across universities. The effect of incentives is mainly to increase the quality rather than the quantity of inventions