Blood pressure is usually measured by conventional sphygmomanometry, a procedure fraught with many potential sources of error. Automated techniques of measurement, such as ambulatory monitoring and self-measurement, reduce the limitations of conventional sphygmomanometry. However, the diagnostic thresholds applicable for conventional sphygmomanometry cannot be extrapolated to automated measurements. During the past 10 years criteria for normality have gradually been developed for ambulatory blood pressure monitoring. First, the distributions of the ambulatory blood pressure in normotensive subjects and in untreated hypertensive patients who had initially been classified on the basis of their conventional blood pressure were studied. Second, epidemiological studies were performed to investigate the distribution of the conventional and ambulatory blood pressures in the population at large. Third, authors of several studies have now validated the preliminary thresholds for ambulatory monitoring against left ventricular hypertrophy, other signs of target-organ damage and the incidence of cardiovascular complications. Finally, authors of clinical trials investigated whether it is beneficial to patients and cost-effective to diagnose and treat hypertension on the basis of ambulatory monitoring rather than under the sole guidance of conventional sphygmomanometry. For systolic/diastolic measurements, the upper limits of normotension include 130/80, 135/85 and 120/70 mmHg for the 24 h, daytime and night-time blood pressures, respectively. Whereas for ambulatory monitoring a large body of evidence currently supports the proposed diagnostic thresholds, for the self-recorded blood pressure, to a large extent, this evidence must still be collected. In pursuing this goal, the methods applied for ambulatory monitoring may serve as a template. On the basis of a meta-analysis of summary statistics of published articles and a meta-analysis of data from individual subjects, 135/85 mmHg is likely to be the upper limit of normality for the self-measured blood pressure. Obviously, this threshold is preliminary and must be further validated in prognostic studies. However, the present proposal could guide clinicians who wish to use self-measurement to refine the diagnosis and the management of hypertension based on conventional sphygmomanometry.