A color prediction model of printer based on radial basis function (RBF) neural network optimized by genetic algorithm (GA) and subspace partition is presented to settle the nonlinear of printer and complexity of printing conditions. The color space of printer is divided into subspaces and the models are built in subspaces, GA-RBF neural network model is built by GA optimizing the hidden layer nodes and width parameters of RBF neural network. Prediction accuracy of the proposed algorithm is compared with RBF neural network and cellar Yule-Nielsen spectral neugebaue (CYNSN) model. Experimental results show that GA makes up for the defect of single adjustable parameter of RBF neural network and improves prediction accuracy. Compared with other models, the proposed model has high prediction accuracy and generalization ability. It is feasible for color prediction of printer.