For patients with hypertension, an individual risk prediction tool for cardiovascular disease based on on-treatment blood pressure is needed and would be useful. The objective of this study was to establish a 3-year risk prediction model for cardiovascular disease based on data from 13 052 patients with no history of cardiovascular disease in the Olmesartan Mega study to determine the relationship between Cardiovascular Endpoints and Blood Pressure Goal Achievement study. To develop dynamic prediction models including on-treatment blood pressure, a Cox proportional hazard model using the sliding landmarking method with three landmark points (6, 12 and 18 months from baseline) was used. The prediction model included blood pressure (<130/85 mm Hg, ⩾130/85 to <140/90 mm Hg, ⩾140/90 to <160/100 mm Hg and ⩾160/100 mm Hg) as a time-dependent covariate and well-known baseline risk factors (sex, age, smoking, family history of coronary artery disease and diabetes) as covariates. The 3-year risk assessment chart was constructed using the combination of all risk factors in the prediction model, and six different colors were displayed on each chart corresponding to the predicted probability of cardiovascular disease. Judging from the chart, if an elderly man with diabetes and other risk factors had a blood pressure of <130/85 mm Hg at 6 months, the risk of cardiovascular disease would be 8.0%, whereas the risk would be 8.6% if he had a blood pressure of ⩾130/85 to <140/90 mm Hg. The risk assessment chart developed from the large-scale observational study data would help physicians to more easily assess the cardiovascular disease risk for hypertensive patients on antihypertensive treatments.