Federal agencies and universities in the U.S. promote interdisciplinary research because it presumably spurs transformative, innovative science. Using data on almost 900 research-center–based scientists and their 32,000 published articles, along with a set of unpublished papers, we assess whether such research is indeed beneficial and whether costs accompany the potential benefits. Existing research highlights this tension: whereas the innovation literature suggests that spanning disciplines is beneficial because it allows scientists to see connections across fields, the categories literature suggests that spanning disciplines is penalized because the resulting research may be lower quality or confusing to place. To investigate this, we empirically distinguish production and reception effects and highlight a new production penalty: lower productivity, which may be attributable to cognitive and collaborative challenges associated with interdisciplinary research and/or hurdles in the review process. Using an innovative measure of interdisciplinary research that considers the similarity of the disciplines spanned, we document both penalties (fewer papers published) and benefits (increased citations) associated with it and show that it is a high-risk, high-reward endeavor, one that partly depends on field-level interdisciplinarity.