White men who responded positively to the Rose questionnaire but who were judged on clinical grounds to be free of angina pectoris resembled true cases of angina pectoris insofar as their blood pressure and cholesterol level was concerned. This was true for results obtained at the time of the current examination, as well as those taken seven years prior to this examination. Variability in response to the questionnaire at an interval of one year was assessed. "True" cases of angina showed as much variation (changing from positive to negative responses) as did the "false positives." This would suggest that, in large part, this variation is a reflection of the natural history of the disease rather than unreliability of the questionnaire. The Rose questionnaire was found to have a sensitivity of 81% and a specificity of 97% when compared to clinical judgment.