Most of the color prediction models use a single dot gain curve for each primary ink. In the proposed model, the behavior of the dot gain of each primary ink is characterized by three curves based on CIEXYZ tristimulus values. In our previous works, it was shown that the usage of three characterization curves for each primary ink reduced the color difference between the predicted and measured data compared with the simple Yule–Nielsen model. In this article, an effective coverage map based on CIEXYZ is created. This map presents the effective coverage values of the primary inks corresponding to different ink combinations. Given any reference ink combination, the effective coverage values of the involved primary inks are estimated by cubic interpolation. Compared to our previous models, the proposed model gives significant reduction in the color difference between the predicted and the measured data.