The one-, two-, and three-dimensional solubility parameter models were used to analyze solution thermodynamic data of several solutes in poly(vinyl chloride) (PVC) and poly(tetramethylene glycol) (polytetrahydrofuran, PTHF). A previous proposed method to estimate solubility parameter, based on direct minimization of the sum of the squares of errors in the prediction of XRTIV, was extended to two- and three-dimensional, and the results of solubility parameter components were the same as the linear regression method. The one-dimensional model gave the smallest sum of the squares of errors, followed by the three-dimensional model, then the two-dimensional model. A systematic error was identified when two- and three-dimensional models were used for the data. A method using less weight on polar and hydrogen-bonding components produced smaller errors for the two- and three-dimensional models. (C) 2004 Elsevier B.V. All rights reserved.