This study tested the hypothesis that a sphygmomanometer cuff bladder long enough to encircle the arm in most adults ('obese cuff) would provide a more accurate and precise estimate of intra-arterial pressure than the usual 'standard ' cuff bladder. In 53 patients undergoing diagnostic coronary angiography (35 males, 18 females, aged 36–79 years), indirect blood pressure, measured in the left arm with a random-zero sphygmomanometer, was compared with simultaneously measured femoral intra-arterial pressure. Duplicate indirect measurements were made with each of two cuffs containing bladders measuring 39 15 cm ('obese') and 23 12 cm ('standard').The obese cuff bladder encircled 80% or more of the arm circumference in all subjects, whereas the standard cuff bladder met this requirement in only 19% of the subjects. For both systolic and diastolic pressure there was marked interindividual variability in the differences between indirect and direct measurements with both cuffs. With the obese cuff there was no systematic error in the diastolic blood pressure measurement. The standard cuff consistently overestimated diastolic pressure by 7.7 ± 8.3mmHg (mean ±s.d.). For both cuffs, the difference between indirect and direct diastolic pressure increased with arm size (P<0.05). Both cuffs underestimated systolic blood pressure, the obese cuff by 15.5 ± 11.7mmHg and the standard cuff by 7.6 ± 12.1 mmHg. These systolic blood pressure underestimates were greater at higher blood pressures (P<0.01) and with smaller arms (P<0.05). Age was not related to measurement error with either cuff.These results have demonstrated the inherent problems with indirect blood pressure measurement by sphygmomanometry, viz. its unreliability and the consistent underestimation of systolic pressure. The obese cuff is more appropriate than the standard cuff for routine adult blood pressure measurements, as it provides a more accurate estimate of diastolic pressure, the accepted criterion for the diagnosis and management of hypertension.