The authors used the diagnosis of craniosynostosis to compare subjective evaluation of image quality with objective diagnostic utility. They studied in detail the responses of one observer, who read plain radiographs, computed tomographic (CT) scans, and three-dimensional reconstructions of CT scans (obtained with three different methods) for 82 patients with this diagnosis. The observer rated image quality and certainty in diagnosis made from each image. Subjective and objective performances were found to be strongly linked. High-quality images served as the basis for more accurate diagnoses than low-quality images. The increase in diagnostic performance results primarily from increased specificity, a fact that suggests that specificity and the concomitant diagnosis of normalcy are the focus of attention when image quality is evaluated.