The accuracy of auscultatory blood pressure (BP) determination was assessed in patients with chronic atrial fibrillation by performing simultaneous auscultatory BP determination on the upper arm and a direct BP determination on the contralateral arm. The subjects were three hospitalized patients, aged from 52 to 75 years. A Teflon catheter was introduced into the radial artery which was connected to a pressure transducer, and a cuff was twisted around the contralateral upper arm in the supine position. Simultaneous recording of directly determined BP and cuff pressure enabled the comparison of direct BP with auscultatory BP. The appearance of the Korotkoff I sound (systolic BP) and V sound (diastolic BP) was marked on the cuff pressure curve. This maneuver was repeated five times in each patient. The method of Bland and Altman was employed to assess the agreement between auscultatory and direct determinations. The auscultatory method estimated BP with differences of -14.3 to +27.3 mmHg in systolic BP and -12.1 to +11.9 mmHg (+/-2SD) in diastolic BP compared with the direct method. The difference in systolic BP between the auscultatory and the direct methods was greater than that in diastolic BP. Thus, there are unacceptable differences in systolic BP between auscultatory and direct methods that can be attributed to BP fluctuations. The auscultatory method in diastolic BP is more accurate than that in systolic BP and may be more useful in the clinical setting.