The invention relates to a reservoir shear wave velocity prediction method based on rock physics. The method comprises the steps that (1) supposing that the bulk modulus, the shear modulus and the density of a destination interval mineral are unknown constants, the error between a longitudinal wave velocity and a measured longitudinal wave velocity is acquired according to equation modeling, and the error between the acquired density and the measured density is calculated; (2) according to the minimum error criterion, the bulk modulus, the shear modulus and the density of a mineral component are inverted through a genetic algorithm; (3) and the shear wave velocity is calculated. According to the step (1), supposing that the elastic modulus of each rock component is an unknown constant in a reservoir logging curve section, the objective function of the inversion of the logging curve can be defined that a superscript M represents a simulation longitudinal wave velocity, O represents a logging observation longitudinal wave velocity, and w Rho and w rho represent normalized weighted coefficients; and the objective function is a nonlinear function of the bulk modulus, the shear modulus and the density of each mineral component. The method can be applied to a sand shale reservoir, and carbonate rock and other complex reservoirs.