This paper introduces a prediction model of the wayside noise level of Shinkansen. The noise indices used in the model are the A-weighted sound exposure level of a train set pass L_AE, equivalent continuous A-weighted sound pressure level L_Aeq,T and maximum A-weighted sound pressure level (slow) of a train set pass L_pA,SMAX. In this method, Shinkansen noise sources are divided into four components, namely, the noise from the lower parts of cars, concrete bridge structure noise, aerodynamic noise from the upper parts of cars and pantograph noise. Each noise component is regarded as a row of discrete point sources, the positions and power levels of which are decided according to the noise component, train velocity, car type, track and structure. As a result, the wayside noise level of Shinkansen can be predicted under various conditions.