Average 24-h blood pressure (BP) is more representative of usual BP than office BP. However, the clinical relevance of 24-h BP in treated hypertensive subjects is incompletely known. Thus, we studied 395 uncomplicated hypertensive subjects (209 men, 53+/-10 years) who were receiving antihypertensive drug therapy from >1 year. All subjects underwent 24-h ambulatory BP monitoring and M-mode echocardiography. Subjects were classified by tertile of the difference between observed and predicted 24-h systolic BP (the latter determined by regressing 24-h systolic BP on office systolic BP): higher-than-predicted (III tertile), around the regression line (II tertile), and lower-that-predicted (I tertile) 24-h BP. Despite similar office BP (144/89, 141/88, and 144/89 mm Hg in the III, II, and I tertile, P = not significant), age, body mass index, and duration of hypertension, left ventricular mass was greater in the subjects with higher-than-predicted 24-h systolic BP (50+/-14 g x m(-2.7)) than in the other two groups (46+/-13 g x m(-2.7) and 42+/-10 g x m(-2.7), both P < .05). The III tertile also showed a more concentric left ventricular geometric pattern (relative wall thickness was 0.42+/-0.08, 0.40+/-0.07, and 0.38+/-0.07 in the III, II, and I tertile, P < .001) and a reduced systolic function at the midwall level (16.8+/-3, 17.7+/-3, and 18.2+/-3, P < .001). In conclusion, treated hypertensive subjects whose 24-h BP is notably higher than one would predict from office BP are more likely to develop left ventricular hypertrophy, a strong adverse prognostic marker. In a sizable subset of treated hypertensive subjects, BP measured in the physician's office underestimates usual BP and its impact on left ventricular structure.