The paper presents a new methodology for evaluating corporate voluntary disclosures in the annual report discussion section. Based on a new dataset of electronic annual reports and a standard text analysis software package, we text-search a large number of annual reports at minimal (marginal) cost. The resulting sample sizes are comparable to those employed in studies based on the AIMR-FAF database. A major advantage of our new scoring system is that it is adaptable to the particular requirements of the research project. We demonstrate the importance of this feature when applying our new disclosure scores to the case of 'prices leading earnings'. While we are unable to find the predicted association with a broadly defined measure of disclosure quality, our results reverse once we focus on a more narrowly defined metric based on forward-looking profit statements.