In recent years, many approaches have been addressed for gamut mapping. Different color spaces and different gamut mapping methods were investigated to improve gamut mapping accuracy or to achieve perceptual pleasing result. The effects of different color spaces (i.e. CIE L*a*b*, CIE L*u*v*, CIECAM97s, Munsell color system, and other non-standard color spaces) on gamut mapping have been addressed in many publications. However, most of these papers are based on gamut mapping in a single color space. Zeng presented a gamut mapping approach using multiple color spaces to solve problems or limitations in gamut mapping in a single color space. By this approach, several color spaces are chosen to perform gamut mapping in different regions to solve some color space problems. In this paper, another approach will be addressed to solve similar color space problems and also to perform some color preference mapping.Although gamut mapping using multiple color spaces solves some gamut mapping problems, it makes implementation more complicated than gamut mapping using a single color space. It requires a gamut mapping object for each color space, and it also requires additional parameters to decide what color space(s) to select and additional function(s) for smooth transaction from one color space to another. In order to apply the approach of gamut mapping in multiple color spaces but still preserve the simplicity of gamut mapping in a single color space, another approach, which is gamut mapping in a composite color space, was developed. A composite color space is defined as a color space derived from the combination of several color spaces with different weightings in different gamut regions. For the gamut mapping in a composite color space to achieve similar result as that in multiple color spaces, a composite color space and one gamut mapping object are created to replace several gamut mapping objects and transaction functions for gamut mapping. The implementation of gamut mapping in a composite color space is exactly the same as that of gamut mapping in a single color space except that a special color space is applied to replace a regular color space. The composite color spaces that were tested are the linear combinations of standard color spaces, such as CIE L*a*b*, CIE L*u*v*, CIECAM97s, etc.Our analysis shows that gamut mapping in a composite color space is not mathematically exactly equivalent to the gamut mapping in multiple color spaces. However, very close gamut mapping results are achieved with properly defined transaction functions. Our experimental showed that constructing printer ICC profiles using gamut mapping under a composite color space achieved similar result as that with gamut mapping under multiple color spaces in correcting blue shift problem for monitor RGB to printer CMYK color conversion.