The Rose Questionnaire has had high specificity and variable sensitivity when compared to physician identification of the symptom complex of angina pectoris. We assessed the accuracy of a supplemented Rose Questionnaire in a series of 198 patients by comparing the Questionnaire to exercise thallium test evidence of coronary artery disease. The Rose diagnosis of angina had 26% sensitivity, 79% specificity, 42% positive predictive value, and 65% negative predictive value. The Rose diagnosis of myocardial infarction had 26% sensitivity and 90% specificity. The Rose diagnosis of angina or infarction yielded a sensitivity of 44%, specificity of 72%, positive predictive value of 67%, and negative predictive value of 50%. Supplemental questions designed to identify atypical ischemic pain led to increased sensitivity of up to 68% that was offset by decreased specificity. While the Questionnaire's sensitivity for coronary disease was greater for women than men (57 vs 40%), the overall accuracy was the same because specificity was lower (63 vs 80%).