Because many types of electronic imaging devices are now available, cross-media color reproduction technology has received widespread attention due to the need to provide accurate color stimuli for different devices. In the case of cross-media color reproduction between a monitor and a printer, RGB has to be converted into a device-independent color space in order to translate the color information between the two devices. Thereafter, gamut mapping is used to compensate for any gamut mismatch and device-independent colors have to be re-converted into output colors such as CMY control values for printing. For color conversion between device colors and device-independent colors, empirical representation using sample measurements is currently widely utilized. In the case of the printer, color samples are uniformly selected in the colorant space, printed as color patches, and then measured. However, because these color samples are not evenly distributed inside the printer gamut, the color conversion error is increased. Accordingly, this article introduces a color-sampling algorithm for a printer to reduce the error in color conversion, and the performance is analyzed via color conversion experiments using three conversion methods, regression, neural network, and interpolation.