The dielectric constant (e) of a polymer represents the capacitance of the polymer relative to a vacuum and is important in many industrial applications. Nevertheless, accurate experimental e values of polymers are often unavailable, and current prediction methods lack universality and suitability. Reported here is the development and testing of a new QSPR (quantitative structure property relationship) correlation of e for polymers. The entire set of 57 polymers was divided into a training set of 45 polymers and a test set of 12 polymers according to Kennard and Stones algorithm. A nine-descriptor model, with squared correlation coefficient (R(2)) of 0.9384 and standard error of estimation (s) of 0.0873, was produced by using the stepwise multilinear regression analysis (MLRA) on the training set. The reliability of the proposed model was further illustrated using various evaluation techniques: leave-one-out cross-validation procedure, randomization tests, and validation through the test set. POLYM. ENG. SCI., 51:2408-2416, 2011. (C) 2011 Society of Plastics Engineers