Abstract: Stereotomography is a new reflection tomography method based on the slopes of locally coherent seismic events. It can provide accurate velocity macromodels for prestack depth migration. We study the implementation methods for picking travel times and slopes using hyperbolic stack in common midpoint gather and common offset gather, representing the discrete velocity model using B-spline function, tracing rays with the numerical solution of ray equation, calculating picked data based ray theory, determining ray parameters, solving the inverse problem, etc., and then provide the algorithm flow of stereotomography. Based on the experiments with Marmousi model, we test and analyze the main factors of stereotomography like the initial velocity model, the amount of picked data, discrete grid scale, smoothing weight for velocity model, and summarize their influences on the results of stereotomography.