The printing industry has quantified metrics for several kinds of image distortions. Each of these metrics refers to a particular artifact, and thereby to a mechanical origin of the artifact. A metric does not assign a perceptual significance to its associated artifact, but such an assessment is needed, e.g., when one wants to adjust a printing process to trade one sort of distortion against another. A way is needed to assess the perceptual significance of each artifact. Accordingly, this paper describes a perception-based evaluation of individual print-quality metrics, by applying a vision model (the Sarnoff JND Vision Model) that is being used successfully to predict digital-video quality. To adapt this model to the printing application, the following steps have been taken. (1) For several printed renditions of a KDY test pattern, sub-images containing individual artifacts are selected. (2) For each of these images, appropriate KDY ImageXpert quality metrics are computed. (3) The Sarnoff model is used to compare these sub-images with their corresponding bitmaps (warped so as to achieve registration). (4) The Sarnoff model values are compared with representative KDY metrics for each sub-image. From the comparison, it was learned that the perceived distortion in black characters on a white background is less than for white characters on a black background. The JND model separates these categories more completely than many of the KDY metrics. It is expected that if both ImageXpert and JND metrics are used in printing applications, the result will be an accurate assessment of the perceptual magnitude of artifacts (from the JND model), also means to correct them (from ImageXpert).