In many cases, the analysis of human movements could not be made rigorously if something affects the contextual, luminous or material environment. SAGA-3DV has been designed to operate in these various and difficult experimental conditions. To reach this fondamental goal, SAGA-3DV uses a two stages stereovision process. During the first, color video images are recorded on professional and numeric video tape recorders. The second is generally made in laboratory; it consists of analysing the images so as to extract, for each camera, the trajectories of color markers put on the body. SAGA-3DV software greatly automates this operation so as to obtain quickly expected 3D trajectories.