A general quantitative structure-property relationship (QSPR) model, which is based on the structural analysis of polymers, was built to predict refractive indices of linear polymers by applying four molecular descriptors: the sum of valence degrees (SVDe), the degree of unsaturation (DU), the relative number of halogen atoms (RNH) and the electrostatic attract or hydrogen bond between the main chains (Q+/-). Different from other existing models, the model paid more on structure analysis and selecting descriptors, which are of significance to the refractive index of polymers. The results from 121 linear polymers showed that the correlation coefficient (R) between predicted and experimental values was 0.964 and the average prediction error was 0.87%. Compared with the existing QSPR models, the proposed model requires only four descriptors, which can be obtained by simple calculation and makes it easier to predict the refractive indices of polymers. (C) 2004 Elsevier Ltd. All rights reserved.