The study objective was to compare blood pressure (BP) measurement by the Hawksley random-zero sphygmomanometer and the standard mercury sphygmomanometer. Comparison of simultaneous 'blind' BP measurements were made using the Hawksley random-zero sphygmomanometer and the standard mercury sphygmomanometer linked by a Y-connector to a single cuff, in the general practice and office environments. Sixty five healthy volunteers and general practice patients, aged between 20 and 50 years (SBP range 82-184 mm Hg, DBP range 38-112 mm Hg), were studied. Each had three blood pressure measurements taken. Mean BPs recorded by the Hawksley random-zero sphygmomanometer were lower than those recorded by the standard mercury sphygmomanometer. The Hawksley random-zero sphygmomanometer underestimated SBP by 1.3 mm Hg (95% CI 0.9-1.8 mm Hg) and DBP by 1.7 mm Hg (95% CI 1.1-2.3 mm Hg). These differences between instruments were independent of BP level both for systolic and diastolic measurements. An overview including this study and six other published reports describing nine studies examining the performance of the Hawksley random-zero sphygmomanometer suggested a similar degree of underestimation for SBP (mean difference 1.35 mm Hg, 95% CI 1.24-1.46 mm Hg). Underestimation of DBP appeared greater (mean difference 2.54 mm Hg, 95% CI 2.43-2.65 mm Hg) but was reduced when two outlying studies were removed from analysis (mean 1.97, 95% CI 1.85-2.09 mm Hg). We conclude that the Hawksley random-zero sphygmomanometer underestimates systolic and diastolic pressure, when compared with the standard mercury sphygmomanometer. However, the degree of underestimation is small and appears consistent across a wide range of blood pressure levels.(ABSTRACT TRUNCATED AT 250 WORDS)