Foundational theories of the legislature disagree about why, or even whether, legislative leaders are powerful, but issues of measurement and causal inference have prevented empirical work from addressing these debates effectively. To make progress, we offer a new dataset on the identities of legislative leaders in all U.S. state legislatures over the past 20 years. Using a difference-in-differences design, we show that, on average, becoming a majority-party leader causes a large increase in contributions from strategic interest groups—an indication that leaders are indeed powerful. Contrary to major theoretical predictions, however, we show that leaders are no more powerful, and possibly less powerful, when legislative polarization increases. Moreover, neither the size of the majority party nor the professionalization of the legislature are associated with how powerful majority-party leaders are. In contrast, we find that majorityparty leaders are more powerful in bigger legislatures, which we argue suggests that a key role for leaders is solving issues of complexity and coordination. The paper thus offers new data and evidence that revises and improves our understanding of legislative politics both in the U.S. and in democratic settings more generally.