The effect of ambient light on linearization of gray scales was studied in two experiments by measuring brightness-discrimination thresholds for stimuli as a function of ambient-light level. The luminance and contrast levels of stimuli were chosen to include typical instances encountered in electronically displayed medical images. Three ambient-light levels were used: 4 and 40 lx, which span the range that we found in well-controlled radiologic reading rooms set up with electronic displays, and 148 lx, which represents the low end of the ambient-light range found in light-box reading rooms. In the first experiment, discriminability of some test stimuli was found to change as ambient light increased. The pattern of results suggests that a single linearization function would be clinically acceptable in a reading room with ambient light ranging from 4 up to, at least, 40 lx. At 148 lx, however, the just-noticeable-difference versus intensity function is different enough to require a gray-scale linearization function other than our standard one, which is based on 4-lx ambient light. We attribute these changes in stimulus discriminability to changes in the displayed image produced by ambient light, not to changes in the visual sensitivity of the observer. In the second experiment, which controlled for the display changes produced by ambient light, we found no change in stimulus discriminability as a function of ambient light.