To examine the accuracy of the Hawksley random zero sphygmomanometer two studies were done with subjects with a wide range of blood pressure. When readings made by one observer on the UK model of the Hawksley sphygmomanometer were compared with readings by two independent observers on separate mercury sphygmomanometers, the Hawksley device underestimated systolic readings by a mean (SD) of 2.0 (2.4) and 0.5 (3.6) mm Hg and diastolic readings by a mean of 3.7 (2.7) and 2.8 (2.9) mm Hg. When readings made on the UK and US models of the Hawksley sphygmomanometer were compared with those made on mercury sphygmomanometers, with observers exchanging devices half way during the experiment, the UK Hawksley device underestimated systolic pressure by a mean of 3.8 (SD 3.5) mm Hg and diastolic blood pressure by 7.5 (3.8) mm Hg; and the US model by 2.6 (3.4) mm Hg for systolic pressure and 6.2 (3.7) mm Hg for diastolic pressure. There was better agreement between two observers using standard sphygmomanometers than between an observer using the Hawksley random zero sphygmomanometer and an observer using a standard sphygmomanometer. Thus, the quantitative aspects of blood pressure in epidemiological and intervention studies in which the Hawksley random zero sphygmomanometer was used need re-evaluation. Moreover, the Hawksley random zero sphygmomanometer, in its present design, should not be used in hypertension research.