Blood pressure (BP) screening is important to identify those at risk of cardiovascular disease, but there has been little data on the appropriate interval of screening. We aimed to evaluate the optimal interval and the best measure for BP re-screening by estimating the long-term, true change variance ('signal') and short-term, within-person variance ('noise'). Study design was a cohort study from 2005 to 2008. Target population was Japanese healthy adults not taking antihypertensive medication at baseline, in a teaching hospital. We measured annually the systolic BP (SBP) and the diastolic BP (DBP), and calculated the pulse pressure (PP) and the mean arterial pressure (MAP). A total of 15 055 individuals (51% male) with a mean age of 49 years had annual check-ups. Short-term coefficient of variation was lowest for MAP at 5.2%, followed by SBP (5.7%) and DBP (5.8%), and highest for PP (12%). After 3 years, the 'signal' of true BP changes of only SBP and MAP equaled the noise of BP measurement; however, it was larger for those with higher initial BPs. SBP or MAP appears to be a better screening measure. The optimal interval should be 3 years or more, with SBP < 130 mm Hg and 2 years for those with SBP ≥ 130 mm Hg.