Aalto et al. recently proposed a model for compressed liquid densities. The model was found more accurate than the Hankinson-Brobst-Thomson (HBT) and Chang-Zhao models. However, the pressure region of the data studied was limited to 200 bar maximum. In this work, the recently developed liquid density model is extended to high pressures. The equation describing the pressure dependence of liquid density is reformulated and the required parameters are optimized using a database containing 7478 data points for 31 pure hydrocarbons; maximum pressure in this data set is 8000 bar. The average absolute deviation (AAD) between these data and the recommended model is 0.4636%. A comparison to the results obtained with the HBT and Chang-Zhao models for the same data set shows that the new model is clearly more accurate in the extended pressure range, as well. The revised model is also tested in predicting liquid densities for mixtures; 84 different combinations of mixing rules are studied. The evaluation of the mixing rules is carried out using two compilations of experimental data: the first one contains 6712 points for 47 binary and two ternary mixtures, and the second 3582 points for 11 methane + alkane mixtures. In addition, the predictions are tested with a data set of 1119 points for other miscellaneous mixtures. No binary interaction parameters are used. With the recommended mixing rules, the AAD percentage is 0.5824% for the first set of data. If one simply adopts the mixing rules recommended for the HBT model, the AAD value for the same data set becomes 0.7482%.