The hypothesis has been advanced that cardiovascular prognosis is related not only to 24-hour mean blood pressure but also to blood pressure variability. Data, however, are inconsistent, and no long-term prognostic study is available. In 2012 individuals randomly selected from the population of Monza (Milan), 24-hour ambulatory blood pressure (Spacelabs 90207) was measured via readings spaced by 20 minutes. Systolic and diastolic blood pressure variability was obtained by calculating the following: (1) the SD of 24-hour, day, and night mean values; (2) the day-night blood pressure difference; and (3) the residual or erratic blood pressure variability (Fourier spectral analysis). Fatal cardiovascular and noncardiovascular events were registered for 148 months. When adjusted for age, sex, 24-hour mean blood pressure, and other risk factors, there was no relationship between the risk of death and 24-hour, day, and night blood pressure SDs. In contrast, the adjusted risk of cardiovascular death was inversely related to day-night diastolic BP difference (beta coefficient=-0.040; P<0.02) and showed a significant positive relationship with residual diastolic blood pressure variability (beta coefficient=0.175; P<0.002). Twenty-four-hour mean blood pressure attenuation of nocturnal hypotension and erratic diastolic blood pressure variability all independently predicted the mortality risk, with the erratic variability being the most important factor. Our data show that the relationship of blood pressure to prognosis is complex and that phenomena other than 24-hour mean values are involved. They also provide the first evidence that short-term erratic components of blood pressure variability play a prognostic role, with their increase being accompanied by an increased cardiovascular risk.