Angina pectoris (AP) as determined by the Rose questionnaire was assessed in nearly 16,000 black and white men and women participating in the Atherosclerosis Risk in communities Study, a population study of cardiovascular disease in four communities. The questionnaire was administered at yearly intervals and estimates of repeatability were made. Validity was assessed indirectly by comparing Rose AP to risk factors, prevalent heart disease, medication use, and thickness of carotid artery walls as measured by B-mode ultrasound. Using kappa statistics for agreement of positive Rose AP determinations taken 1 year apart, white men show a higher level of agreement than white women (average kappa 0.36 for white men, 0.30 for white women), and whites show a higher level of agreement than blacks (average kappa 0.23 and 0.22 for black men and women, respectively). Rose AP that persists for more than one determination is associated with thicker carotid artery walls, greater amounts of cigarette smoking, greater prevalence of reported heart attack, and greater use of chest pain medications. A single determination of severe Rose AP is also associated with thicker carotid artery walls. These data suggest that multiple reports and the more severe grading of Rose AP (pain reported while walking on the level) are likely to indicate more severe disease; however, a single report using the Rose questionnaire appears valid, i.e., moderately associated with disease and risk factors, and appropriate for use in epidemiological studies.