This paper investigates the use of adaptive neural network techniques for modeling and automatic control of mean arterial pressure through the intravenous infusion of sodium nitroprusside. An indirect model reference-based adaptive nonlinear control scheme with neural networks approximating the unknown nonlinearities is developed. In this formulation nonlinear estimators are used to adaptively approximate the system uncertainty and augment the linear control law for improved performance. The overall design is based on self-tuning the controller to the specific response characteristics of individual patients. Computer simulations illustrate the ability of radial basis function (RBF) networks to model the unknown nonlinearities and improve the closed-loop system characteristics.