Criteria and methodology have been proposed for measuring the validity of medical history questions in the detection of disease states. These methods have been applied to the evaluation of sensitivity and specificity for each of 20 questions from an automated medical history questionnaire used by the Lahey Clinic Foundation. Values of both of these factors have been noted to vary over wide ranges. In a comparison of results for individual questions, the queries on dyspnea occurring after climbing one flight of stairs and chest pain appear to be better screening tests than corresponding questions on exertional dyspnea and substernal pain. The results also suggest possible clues to the improvement of questions of marginal validity. Because of extremely low sensitivity or specificity values, certain questions appear to have only limited usefulness in screening for disease states.