A device in which an offset is added to a focusing error signal to control an objective lens and thereby reduce the generation of cross-talk when a beam spot that is converged on an optical disk crosses over a wall of a groove of a track of the disk. By adding the offset to the focusing error signal, the objective lens is moved out of focus with the disk. Thus, the shape of a beam spot formed on a photo sensor is changed from a circle to an ellipse. A condenser lens provided for forming the beam spot is moved along the optical axis thereof in such a manner that an output of a recorded signal of the disk is at a maximum value. As a result, the objective lens is brought into focus on the disk. Thereafter, the objective lens is focused on the disk in accordance with the focusing error signal to which the offset is added, while maintaining the beam spot in the shape of an ellipse.