The invention relates to a method for stabilizing a video recording of a scene made by a video camera and represented by video data, said method comprising the steps of subdividing said video data into a plurality of successive frames themselves divided into a plurality of blocks, determining for each block of each frame a motion vector representing the direction and magnitude of the motion in said block, said vector at an instant t being called global motion vector GMV(t) and representing said motion at the instant t with respect to the previous frame, defining a modified vector called integrated motion vector IMV(t) at the instant t and designating the final motion vector correction to be applied to the current frame in view of its motion correction, said integrated motion vector being given by the expression:IMV(t) = GMV(t)+a(E). WIV(t-1) where a(E) is a variable adaptive factor depending on an expression E and IMV(t-1) is the integrated motion vector corresponding to the previous current frame, and modifying the video data according to the modified integrated motion vectors defined for each successive current frame.