Several authors have attempted to use evoked potential recording as a means of objective perimetry. This note describes two limitations. These are: 1. (1) The topological distribution of both flicker and pattern EPs over the scalp depends on which retinal site is stimulated. Thus, in normal subjects, EPs can erroneously suggest that one part of the visual field is defective at electrode A while erroneously suggesting at electrode B that a different part of the visual field is defective. 2. (2) Plots of EP amplitude vs. flicker frequency are different for different retinal quadrants. Thus, in normal subjects, flicker EPs can erroneously suggest a field defect in quadrant A at one flicker frequency while erroneously suggesting a field defect in quadrant B at a different flicker frequency. Further complications are that flicker EPs for individual quadrants can differ markedly between subjects and between left and right eyes of individual subjects (though these differences seem less for fullfield stimulation). These two points limit EP perimetry in the sense that quite misleading conclusions can be drawn from either transient or flicker EPs from a single electrode or flicker EPs at a single flicker frequency. It is necessary at least to compare records from several electrode sites and at several flicker frequencies; in itself a laborious procedure unless a computer is used. Even then, the development of an appropriate computer programme would not be trivial work, because the appropriate criteria for abnormality remain to be empirically established by recording from an adequately large population of control subjects.