A colorimetrically characterized computer-controlled CRT display was used to determine closest perceptual color matches of 25 colors when an exact match was not allowed. An artificial but realistic color gamut was created by intersecting the display gamut with a gamut of a Xerox 4920 color laser printer. Each of 21 observers performed color matches between out-of-gamut colors and those on the artificial gamut＇s edge. Each observer made color matches on 4 different images. The images represented some of the categories that business graphic images can fall into. Between the different image types, there ware no multidimensional (MANOVA) statistically significant differences at the 10% confidence level in any of the 25 colors tested. The mapping vectors showed that (1) observers don＇t make simple matches as assumed by most gamut-mapping experiments done to date, (2) the influence of image content for simple graphical images tested does not have a large effect when the task is to make closest perceptual color matches, and (3) CIELAB hue angle is not uniform enough, especially in blue and cyan regions, to make adequate gamut-mapping transforms. A simple model for clipping type gamut mapping is proposed. Results are clipping type gamut mapping is proposed. Results are compared to predictions of a new gamut-mapping technique that minimizes weighted color difference between the target color and the gamut boundary.