The caudal fastigial nucleus (cFN) plays a crucial role in the ability to foveate a static or moving visual target. Its unilateral pharmacological inactivation impairs the horizontal velocity of pursuit eye movements and the horizontal amplitude of saccades. After cFN inactivation, more endpoint variability than normal saccades has been reported and interpreted as an altered planning and/or signal dependent motor noise (Eggert et al. 2015). From the demonstration that larger current enhances the size and velocity of saccades evoked by cFN microstimulation (Quinet & Goffart 2015), we propose that the variability of saccade endpoints actually reflects the variable size of the active population. Indeed, if the muscimol injection is not exactly centered in cFN, the number of neurons which are inactivated should increase as the inhibitory drug diffuses, resulting in a dysmetria that increases with time. We will show examples of cFN inactivation experiments performed in head restrained monkeys where the size of dysmetria does not change with time, and other examples where it increases with time. Thus, the variability of saccade amplitude seems to be the consequence of the number of cFN neurons which are involved in saccade execution. More generally, our observations support the hypothesis that the cerebellar control of the ability of foveate and pursue a visual target consists of adjusting the number of active neurons in the cFN and their firing rate.