Color reproduction is a science in constant development. In this article, a new model to solve the color recipe prediction problem using a genetic algorithm is proposed. The objective is to optimize the color recipe prediction stage by determining the dyes to use in a mixture and their respective proportions to reproduce the target color. Two ranges of dyes were used for dyeing 100% cotton woven fabrics: three reactive dyes (CI Reactive Red 238, CI Reactive Yellow 145, and CI Reactive Blue 235) and four direct dyes (CI Direct Orange 34, CI Direct Red 227, CI Direct Blue 85, and CI Direct Black 22). The criterion of optimization, in reproducing the desired shades, is to minimize the CMC color difference between the desired reference color and the color resulting of the predicted recipe. The proposed algorithm revealed good results with small CMC color differences between target and reproduced colors. The effectiveness of the algorithm was also evaluated and proven by calculating errors between the predicted concentrations in the proposed recipes and the actual concentrations.