During the last decades, small head-mounted video eye trackers have been developed in order to record eye movements. Real-time systems-with a low sampling frequency of 50/60 Hz-are used for clinical vestibular practice, but are generally considered not to be suited for measuring fast eye movements. In this paper, it is shown that saccadic eye movements, having an amplitude of at least 5 degrees, can, in good approximation, be considered to be bandwidth limited up to a frequency of 25-30 Hz. Using the Nyquist theorem to reconstruct saccadic eye movement signals at higher temporal resolutions, it is shown that accurate values for saccade peak velocities, recorded at 50 Hz, can be obtained, but saccade peak accelerations and decelerations cannot. In conclusion, video eye trackers sampling at 50/60 Hz are appropriate for detecting the clinical relevant saccade peak velocities in contrast to what has been stated up till now.