Preliminary evidence suggests that the morning surge (MS) in blood pressure (BP) is an independent predictor of cerebrovascular disease. However, the optimal definition of MS is uncertain. To compare the reproducibility of several MS definitions used in the literature, 132 untreated hypertensives were assessed with ambulatory BP monitoring twice, 2 weeks apart. Five MS definitions were compared. MS-1: the average BP of the first hour after rising minus the average BP of the first hour before rising; MS-2: BP 2 h after rising minus that of 2 h before rising; MS-3: BP 3 h after rising minus that of 3 h before rising; MS-4: BP 2 h after rising minus the average BP during sleep; MS-5: BP 2 h after rising minus the average BP of 3 consecutive readings, centered on the lowest reading during sleep. The reproducibility of each MS definition was assessed using the concordance correlation coefficient (CCC), the standard deviation of differences (SDD) and the coefficient of variation (CV) between repeated MS assessments, and the agreement in detecting "surgers," defined as subjects at the top quartile (Q4) of the MS distribution. CCCs were 0.20/0.30, 0.43/0.45, 0.53/0.51, 0.51/0.47, and 0.46/0.48 (systolic/diastolic) for MS-1 to MS-5 respectively; SDDs were 14.3/11.4, 12.1/9.9, 11.2/9.5, 10.3/8.2, and 11.9/9.8, respectively; CVs were 0.49/0.57, 0.44/0.39, 0.37/0.35, 0.36/0.31, and 0.27/0.24, respectively; and the agreement in detecting "surgers" was 69%/70%, 71%/76%, 75%/75%, 81%/83%, and 74%/75%, with kappa of 0.18/0.20, 0.23/0.36, 0.33/0.33, 0.49/0.53 and 0.29/0.31, respectively. There are important differences in the reproducibility of MS calculated by different methods. MS4 appears to provide the most reproducible definition of MS.