The objective of this study was to elucidate the long-term prognostic significance of ambulatory blood pressure. Ambulatory and casual blood pressure values were obtained from 1332 subjects (872 women and 460 men) aged >or=40 years from the general population of a rural Japanese community. Survival was then followed for 14 370 patient years and analyzed by a Cox hazard model adjusted for possible confounding factors. There were 72 cardiovascular deaths during the 10.8-year follow-up. The relationship between 24-hour systolic blood pressure and the cardiovascular mortality risk was U-shaped in the first 5 years, then changed to J-shaped over the rest of the 10.8-year follow-up. After censoring the first 2 years of data, the risk flattened until it again increased for the fifth quintile of 24-hour systolic blood pressure for the 10.8-year follow-up period. For 24-hour diastolic blood pressure, the J-shaped relationship remained unchanged, regardless of follow-up duration and censoring. Ambulatory systolic blood pressure values consistently showed stronger predictive power for cardiovascular mortality risk than did casual systolic blood pressure in the 10.8-year follow-up data, whereas such relationships became more marked after censoring the first 2 years. When nighttime and daytime systolic blood pressure values were simultaneously included in the same Cox model, only nighttime blood pressure significantly predicted the cardiovascular mortality risk for the 10.8-year follow-up data. We conclude that the relationship between ambulatory systolic blood pressure and cardiovascular mortality is not U-shaped or J-shaped, and that nighttime blood pressure has better prognostic value than daytime blood pressure.