Monitoring ambulatory blood pressure, instead of taking pressure readings in hospital, avoids the so-called white-coat effect and allows more readings to be obtained over a longer period of time. It improves the accuracy of the blood pressure estimate and increases the statistical power of therapeutic trials for hypertension. Subjects with white-coat or office hypertension can be detected by ambulatory blood pressure monitoring and excluded from clinical trials. In 23 studies, including a total of 3304 normotensive subjects, the 24-h ambulatory blood pressure averaged 118/72 mmHg; the daytime and night-time pressures were 123/76 mmHg and 106/64 mmHg, respectively. If the mean plus two standard deviation (s.d.) interval is considered the upper limit of normal, the meta-analysis suggested that hypertension may be suspected if the 24-h pressure exceeds 129/87 mmHg, or if the daytime or night-time pressures are higher than 146/91 mmHg or 127/79 mmHg, respectively. On balance, most studies suggest that placebo effects on blood pressure are not observed when blood pressure is measured with ambulatory recorders. If confirmed, this observation indicates that it is possible to simplify the design of trials in the field of hypertension. Ambulatory blood pressure readings should be obtained with properly validated monitors. If the recordings are of sufficient quality, editing does not increase the precision of the subsequent statistical analyses. The statistical analyses should account for diurnal rhythms, and subject and treatment effects.