The Hawksley random zero sphygmomanometer was designed to eliminate observer bias and two digit preference. We have measured blood pressure (BP) in a group of 62 young adults (median age 26.1 years, range 20.2-31.3 years) with reflux nephropathy under standardised conditions (that is, in the morning, after a 2 h supine rest, before venepuncture, using a standard 12 x 23 cm adult size cuff appropriate for the machine used) utilising the random zero sphygmomanometer and the automatic oscillometric BP monitor (Dinamap 8100, Critikon). Seven consecutive recordings of right brachial BP at intervals of 2 min were taken using each instrument alternatively, and the first reading was discarded. The first instrument used to measure BP was alternated between patients to eliminate bias on instrument preference. Random zero sphygmomanometer was used as recommended by the manufacturers and Korotkoff phase V was used to measure the DBP. The observer and the equipment used were the same throughout the study period. The mean SBP and DBP were calculated to the nearest 1 mm Hg utilising the three recordings taken by each instrument. The limits of agreement and the repeatability coefficients for each method of measurement were assessed utilising the statistical method described by Bland and Altman in 1986. The correlation coefficients among random zero and automatic oscillometric BP monitor for SBP and DBP measurements were 0.84 and 0.67, respectively. The average BP (mean of random zero and automatic oscillometric BP monitor measurement) plotted against the difference between the two methods of measurement showed no relation between the difference and the average of measurement in the ranges of BP studied, that is, between 100 and 160 mm Hg systolic and 55 and 100 mm Hg diastolic. The mean of difference between random zero and automatic oscillometric BP monitor for SBP was -6.45 (s.d. 6.07) and for DBP +10.77 (s.d. 8.16) mm Hg. The limit of agreement for SBP measurement was +5.69 to -18.59 mm Hg and for DBP was +27.09 to -5.55. The repeatability coefficients of random zero and automatic oscillometric BP monitor for systolic and diastolic measurements were 8.64 and 7.04, and 9.72 and 6.62, respectively. Bland and Altman analysis indicates major differences between the two methods of measurement. The automatic oscillometric BP monitor could on average over-read the systolic by 6.45 mm Hg and under-read the diastolic by 10.77 mm Hg compared with that of random zero. Furthermore, the limits of agreement were wide enough for a normotensive to be inadvertently defined as a hypertensive on machine error alone. This clearly indicates that automatic oscillometric BP monitor and random zero BP measurements cannot be used interchangeably in clinical practice. Furthermore, the repeatability coefficients, which should ideally be zero, are too large for either instrument to be considered as the gold standard for BP measurement, although that of automatic oscillometric BP monitor was superior to that of random zero. This study highlights the importance of using nomograms generated by the same method of measurement for comparison both in paediatric and adult practice for correct interpretation of BP.