A satisfactory model is developed using CODESSA PRO for the correlation and prediction of milk to plasma concentration ratios (M/P ratio) for diverse pharmaceuticals. A set of experimentally derived MIP ratio values were collected from the literature for 115 widely used pharmaceuticals. The experimental logarithmic MIP ratios were tested with more than 850 theoretical molecular descriptors including constitutional, topological, geometrical, quantum chemical, thermodynamic, and electrostatic types. Based on the data set, for 100 commonly used drugs, a seven-parameter QSAR model was derived that shows a satisfactory (R-2 = 0.791) correlation between predicted and observed values of log(MIP) ratio. (C) 2005 Elsevier Ltd. All rights reserved.