This paper proposes an adaptive neuro-fuzzy model to study the pathophysiology of essential hypertension. Using diverse inputs such as risk factors, physical relations and medical interventions, and states that include both transient and resting states for key physiological variables (blood pressure, total peripheral resistance), it can roughly predict both real-time and long-term blood pressure change for a robust range of inputs. Although it was tuned using published population data, it can be applied to specific individuals to estimate the risks of hypertension with different life experience.