This paper demonstrates a modeling technique of prosthetic heart valves. In the modeling, a pumping cycle is divided into four phases, in which the state of the valve and flow is different. The pressure-flow relation across the valve is formulated separately in each phase. This technique is developed to build a mathematical model used in the real time estimation of the hemodynamic state under artificial heart pumping. The model built by this technique is simple enough for saving the computational time in the real time estimation. The model is described by the first-order ordinary differential equation with 12 parameters. These parameters can be uniquely determined beforehand from in-vitro experimental data. It is shown that the model can adapt, with sufficient accuracy, to a change in the practical pumping condition and the viscosity of the fluid in their practical range, and is also demonstrated that the estimated backflow volume by model agrees closely with the actual one.