Blood pressure (BP) recordings were obtained every 30 min using a noninvasive ambulatory BP recorder during 24 hours in 93 subjects with hypertension. Recordings were analyzed in four different situations: (1) annual check up at the health service center, (2) at work, (3) at home, and (4) during sleep. Subjects with BP higher than 160/95 mmHg in any of the three situations were defined as established hypertensive (n = 36); Subjects with 140-159/90-94 mmHg, as the borderline hypertensive (n = 32); and subjects with BP lower than 139/89 mmHg, as the normotensive (n = 25). Four successive systolic and diastolic BP values were averaged and served as 24-hour BP patterns. Systolic BP over 140 mmHg, diastolic BP over 90 mmHg and mean BP over 107 mmHg were defined as a s-hyperbaric index (s-HBI), d-HBI and m-HBI, respectively. The normotensive group showed a marked decrease of BP after recordings explained by the "white coat phenomenon". The 24-hour recordings in all the three groups showed the highest BP at work and the lowest during sleep. The situational BP changes were generally similar, but the established hypertensive group showed larger BP differences between work and home (SBP 8.8 +/- 9.0 mmHg, 7.8 +/- 8.3 mmHg). For differentiating the three groups, the hyperbaric index was better than the 24-hour average BP. The average home BP showed a good correlation with the 24-hour average BP [SBP: r = 0.94 (p less than 0.01), DBP: r = 0.85 (p less than 0.01)]. Especially, BPs recorded before sleep predicted well the 24-hour average BP. An average of multiple BP measurements before sleep is thought to represent a 24-hour average BP more closely than a conventionally used single recording at home or after awakening.