The muddle of models: what you don't know can hurt you

Prior to clinical trials, animal models remain the primary tools by which the pharmaceutical industry demonstrates the efficacy and safety of candidate therapeutic agents. Despite this reliance, much remains unknown about the models that we use and the factors that affect how they perform. This ignorance can lead to inappropriate assumptions, questionable procedures, and incorrect conclusions from data generated in these models. This perspective provides a narrative of several such instances and provides recommendations as to how we might address our knowledge gap by better characterization of our models, better understanding of our practices, and better control of our studies.