BACKGROUND: In order to determine reference values for ambulatory blood pressure, a random population sample of 1057 persons, 20-88 years old, was investigated in a geographically defined area of Belgium. This article is the final report on the cross-sectional phase of this population survey. METHODS:Twenty-four-hour ambulatory pressure was recorded at 20 min intervals from 0800 to 2200 h and at 45 min intervals from 2200 to 0800 h. Conventional blood pressure was measured by trained nurses at the participants' homes and also in a subgroup of 532 persons at a locally organized clinic. A conventional blood pressure exceeding 140 mmHg systolic or 90 mmHg diastolic and the taking of antihypertensive drugs were the criteria used to distinguish between normotensive and hypertensive persons. RESULTS: In the 1057 patients, of whom 328 were hypertensive, 24 h, daytime (2200 to 0800 h) and night-time (0000 to 0600 h) pressures averaged 119/71, 125/77 and 108/62 mmHg, respectively. Compared with daytime values, blood pressures at home were 3.5/1.5 mmHg lower in 729 normotensive people but 11.6/4.5 mmHg higher in 328 hypertensive patients. In the normotensive subgroup the 95th percentiles of the 24 h, daytime and night-time pressures were 129/80, 137/88 and 121/72 mmHg, respectively. These boundaries were not materially altered when we considered only the 275 participants who had been normotensive both at home and at the clinic (127/79, 135/87 and 118/72 mmHg, respectively). When, in addition to the Belgian data, other reports on large cohorts were also analysed, the transition from normotension to hypertension on ambulatory measurement was likely to be within the ranges of 130-135/80-85, 135-140/85-90 and 120-125/70-75 mmHg for 24 h, daytime and night-time pressures, respectively. CONCLUSION: In comparison with other population surveys and with the earlier interim reports on the Belgian study, the present analysis produced remarkably consistent results with respect to the distributions of the ambulatory measurements. The working definitions of normality based on the 95th percentiles of the ambulatory measurements in the normotensive participants in the present survey and various other studies need further validation in terms of the incidence of cardiovascular complications. For this purpose, the Belgian participants as well as other cohorts are being prospectively followed.