Over the past decade, the US National Institutes of Health has made a series of reforms aimed at increasing the rigor and reproducibility of the research it funds. These changes happened with surprising speed—the NIH Office of the Director announced its plans for reforms in 2014, and by early 2016 these changes had been rolled out across all institutes and centers. What were the historical circumstances that laid the groundwork for these rapid policy changes? This talk argues that the evidence-based medicine movement and declining productivity in the pharmaceutical industry created the conditions of possibility for the emergence of a crisis in preclinical research. For most of the 2000s, the received wisdom was that slowdowns in drug development were the result of a lack of translational research infrastructure that could move knowledge from bench to bedside. A series of papers in late 2011 and early 2012 flipped this narrative, arguing that the true problem was the quality of the knowledge being produced at the bench. What seemed to many to be an implausible argument found support from researchers who had been applying EMB techniques to preclinical data, collecting evidence of bias in bench research. Identifying these origins of this contemporary crisis is helpful for thinking about who is most likely to benefit from reproducibility reforms.