The breadths of the dislocation images, found by x-ray topography of nearly perfect crystals, are explained by two simple models. In the first model the breadth is determined by a critical value of the reflecting plane curvature near the dislocation line. In the second model a critical value of the reflecting plane disorientation is used. Both models are compatible with experimental values of image breadths. As it can be shown, the kinematical theory of disolation contrast in electron transmission microscopy 11 is in better agrement with the first model.