The accurate colorimetric characterization of liquid crystal displays (LCDs) has many difficulties, such as channel interaction and non-constancy of channel chromaticity. Based on the well-performed masking model, a new colorimetric characterization method, namely, sub-space compensation model (SC model) concerning the advantages of the sub-space partition is proposed for LCDs. Firstly, the training data corresponding to different RGB subspaces should be measured, then the relationship between the prediction error of XYZ tristimulus values and the input drive data can be calculated using quadratic polynomial, so that the XYZ prediction error of masking model will be compensated. For the four tested LCDs, the experimental results show that the best mean prediction color differences of the SC model are 0.2783, 0.3199, 0.7090 and 1.2216 CIELAB unit, respectively, which verify that the prediction accuracy of SC model outperforms other conventional models, and that the new model can be applied to all kinds of LCDs.