In the earth observation domain, two classes of sensors may be distinguished: a class for which sensor performances are driven by radiometric accuracy of the images and a class for which sensor performances are driven by spatial resolution. In this latter case, as spatial resolution depends on the triplet constituted by the Ground Sampling Distance (GSD), Modulation Transfer Function (MTF), and Signal to Noise Ratio (SNR), refocusing, acting as an MTF improvement, is very important. Refocusing is not difficult by itself as far as the on-board mechanism is reliable. The difficulty is on the defocus assessment side. Some methods such as those used for the SPOT family rely on the ability of the satellite to image the same landscape with two focusing positions. This can be done with a bi-sensor configuration, with adequate focal plane, or with the satellite agility. A new generation of refocusing mechanism will be taken aboard Pleiades. As the speed of this mechanism will be much slower than the speed of the older generation, it won't be possible, despite the agility of the satellite, to image the same landscape with two focusing positions on the same orbit. That's why methods relying on MTF measurement with edge method have been studied. This paper describes the methods and the work done to assess the defocus measurement accuracy in the Pleiades context.