We utilized a diverse set of drug compounds in order to quantitatively model and predict the drug release profiles of the compounds from starch acetate matrix tablets. The set of drug molecules used represented a wide spectrum of physicochemically different drugs which are administered perorally. The molecular descriptors that characterize the properties of dissolved drugs were evaluated as variables describing the dissolution profiles. The molecular descriptors revealed potential in predicting the dissolution profiles of test drugs, e.g., the molecular level interactions between drug, solvent, and excipient. This method could be utilized as a process control or a development tool in the pharmaceutical industry, e.g., to provide information about drug dissolution behavior in early formulation studies and in testing the similarity of tablets in different batches. Copyright © 2008 John Wiley & Sons, Ltd.