Impact of blood pressure (BP) visit-to-visit variability remains controversial for untreated hypertensives and for normotensive subjects. Association between 6-year systolic BP change and all-cause and cardiovascular (CVD) mortality was studied in general primary care population including untreated hypertensive and normotensive subjects. Normotensive individuals and untreated high BP patients (40,926 and 14,283, respectively) had two check-ups (interval: 5.8±2.2 years) at the IPC center (Paris). Follow-up was 6.1±3.2 years: 1131 people died, 114 from CVD causes. Systolic BP (SBP) change was assessed by tertiles of absolute differences between V1 and V2, and the relationship with mortality was analysed with multivariate Cox models (hazard ratio (HR), 95% confidence interval (CI)) including V1 or V2 parameters notably SBP. Adjusting for V1 variables, mortality was associated with SBP change, for the entire population (all-cause: HR=1.15 (95%: 1.01-1.30) and CVD: HR=1.95 (95%: 1.25-3.05)) and in hypertensive individuals: (HR=1.31 (95%: 1.08-1.59) and HR=2.51 (95%: 1.34-4.72), respectively). Adjustments for V2 variables gave similar results. For those who were normotensive at V1, the associations were not significant. In this primary care population, individual long-term visit-to-visit change of BP is an independent predictor of all-cause mortality in hypertensive individuals but not in those with normal BP.