Engineers often demand generalized models without sophisticated and long-time computations. To date, such models are still lacking for the density prediction of ionic liquid (IL) mixtures. In this paper, corresponding states principle combining with new mixing rules is employed to develop two new generalized models for density prediction of IL mixtures, including an extended Riedel (ER) model and an artificial neural network (ANN) model. A total of 1985 data points of binary and ternary mixtures of IL with molecular solvents, such as water, alcohols, ketones, ethers, hydrocarbons, esters, and acetonitrile, are used to verify the models. Average absolute relative deviations of the ER model and the ANN model are 0.92% and 0.37%, respectively, which indicates both the developed models can achieve a universal and accurate density prediction of IL mixtures. Moreover, the ER model does not contain any fitted parameters and thus provides a real predictive method.