The purpose of this study was to propose reference values, from a viewpoint of prognostic significance, for blood pressure (BP) measured at home with a semiautomated device (home BP measurement) to differentiate normotension and hypertension. We obtained home BP measurements for 1,913 population-based subjects aged 40 years and over in a rural Japanese community and followed up their survival for a mean duration of 5.0 years. There were 141 deaths during the follow-up period. The association between baseline BP values and the overall mortality was examined by Cox proportional hazards regression model, adjusted for age, gender, and the use of antihypertensive medication. The results indicated that the predictive power of home BP level for subsequent mortality was stronger than that of casual screening BP. There was a linear association between home systolic BP and mortality. The association between home diastolic BP and mortality was nonlinear and well approximated with the secondary degree equation of diastolic BP values. Based on this relation, we propose that the reference value for hypertension is 137/84 mm Hg, and normotension is below 137 mm Hg for home systolic BP and between 66 and 83 mm Hg for home diastolic BP. Home diastolic BP below 66 mm Hg should be considered as low diastolic blood pressure. In this population, home systolic BP of 137 mm Hg and home diastolic BP of 84 mm Hg corresponded to the 80th and 87th percentiles, respectively. Then, 29% of the subjects were classified as having hypertension, 52% as normotension, and 19% as low diastolic blood pressure. All previous studies proposing reference values for home BP measurement, derived from cross-sectional observations, were based on the statistical distribution of home BP values. The reference value must, however, be the one that best predicts the risk for morbidity and mortality from hypertension-related complications. This is the first report proposing reference values for home BP measurement based on prognostic criteria.