Identifying patients at high risk for cardiovascular disease is the cornerstone of primary prevention in clinical practice. Early detection of patients at risk enables physicians to treat patients early and possibly prevent CV-events. The Framingham risk score is one of the best validated risk algorithms currently available and has showed excellent discriminative power for the traditional risk factors including male sex, age, systolic blood pressure, antihypertensive treatment, total and HDL-cholesterol, smoking, and diabetes mellitus in predicting CV-events. Since a 10-year risk score might underestimate the true cardiovascular disease burden, especially in younger individuals and women, longer risk estimates might provide a more accurate risk score. In this paper the authors present 30-year follow-up data of the Framingham offspring cohort. Participants were followed for a maximum of 35 years (median, 32 year), with constant monitoring of CV events and mortality, including 671 participants (219 women) with a first CV event, and 622 (267 women) who died of non-CVD causes. CV-events were taken as the primary outcome of interest and defined as a composite of CVD (coronary death, myocardial infarction) and stroke (fatal and nonfatal). The 30-year Kaplan-Meier rate of CVD adjusted for the competing risk of non-CVD death was 7.6% for women and 18.3% for men. In a predictive model, the effect of risk factors measured at baseline on the long-term (30-year) risk of CVD using Cox regression, showed that standard CVD risk factors (male sex, age, systolic blood pressure, antihypertensive treatment, total and HDL-cholesterol, smoking, and diabetes mellitus) were highly significant (p=0.01) in a multivariate model. Diastolic blood pressure and triglycerides were not statistically significant, and inclusion of LDL-cholesterol in place of total cholesterol did not improve model performance. Comparison between 10-year risk scores and 30-year CV-event rates clearly indicate that different combinations of 10-year risks cannot adequately replace 30-year risk scores. For example, 10-year risk for a 25-year-old smoking woman with adverse lipid profile and hypertension is only 1.4%, but the corresponding 30-year risk reaches 12%. Although promising, long-term follow-up does not take into account any changes in participants after the baseline measurements. For example, subjects who stopped smoking after baseline measurement will be defined as smoker for the entire study period and might therefore be misclassified. Nonetheless, accurate long term risk prediction, especially for women, young individuals and subjects in an intermediate risk range, should be assessed according to risk algorithms with a prolonged follow-up to more accurately determine their risk for future CVD. John JP Kastelein, MD