The aim of the study was to determine the relations of 24-h blood pressure (BP) and its different phases with left atrial size. A total of 130 subjects (mean age 46 years) not taking cardiac drugs were studied by M-mode and Doppler echocardiography and ambulatory BP recording. Subjects (excluding those with coronary artery or valvular heart disease, heart failure, or diabetes) were classified into two groups: 25 normotensives and 105 hypertensives (history of antihypertensive treatment and office diastolic BP > 90 mm Hg). The two groups were comparable in terms of sex, age, and heart rate, whereas body mass index, (P < .01), office BP, average 24-h BP, and average daytime and nighttime BP (all P < .00001) were higher in hypertensives. Hypertensives also had increased left atrial dimension, left atrial dimension/aortic root ratio (both P < .001), and left ventricular mass (LV) indexed for height (P < .0001). Positive correlations of left atrial dimension were found with office BP, average 24-h, average daytime and nighttime systolic and diastolic BP, LV mass index, and Doppler-derived E/A ratio. In a multivariate model that included potentially confounding factors, only body mass index (standardized beta coefficient = 0.41, P < .00001), average nighttime diastolic BP (beta = 0.33, P < .00001), and male sex (beta = 0.18, P < .01) were independent predictors of left atrial size in the pooled population. In conclusion, left atrial size is more closely related to ambulatory, rather than office, BP measurements, and high average nighttime BP is a powerful marker of left atrial enlargement in arterial hypertension.