In this issue, Cagnon and colleagues (1) present their quality control program for the American College of Radiology Imaging Network (ACRIN) part of the National Lung Screen Trial (NLST). For the ACRIN-NLST study, more than 18,000 participants were screened for lung cancer with radiography and computed tomography. There were many different imaging centers, models of multidetector helical CT scanners, and models of digital, computed and film radiography equipment. The role of the program described goes well beyond what we would ordinarily think of as “quality control.” Scientifically, it is the heart and soul of the trial, particularly because there seems to have been little consensus prior to the trial on an appropriate low-dose CT lung screening protocol. Without a clear definition of the imaging that could apply across the entire study and ways of checking whether that definition was applied, it is unlikely that a trial with so many patients, radiologists, sites, and scanners would succeed in answering the question of whether CT screening for lung cancer is better than radiography. The judgments that the physicists made to formulate this definition (e.g., how to “strike a balance between image quality and ionizing radiation”) will determine the usefulness of the conclusions from the ACRIN trial. Although the paper may leave some readers hungry for more detail about protocol compliance, it does suggest that, despite extensive measures, half of all sites required corrective intervention but that the problems were quickly and effectively settled. The large, multi-center, randomized, controlled trials provided by ACRIN are impressive in their ability to marshal large amounts of data to address important questions. That said, it must also be acknowledged that most of our scientific knowledge about medical imaging does