OBJECTIVE Diagnosis of functional visual field loss, that is, field loss lacking objective corollaries, has long relied on kinetic visual field examinations using tangent screens or manual perimeters. The modern dominance of automated static perimeters requires the formulation of new diagnostic criteria. METHODS Retrospective review of automated perimetry records from 36 subjects meeting clinical and tangent screen criteria for functional visual field loss. Thirty-three normal eyes and 57 eyes with true lesions, including optic nerve compression, glaucoma, anterior ischaemic optic neuropathy and vigabatrin toxicity, served as controls. RESULTS Standard automated perimetry statistics were unable to reliably discriminate organic versus non-organic visual field loss. Subjective evaluation of perimetric maps indicated that functional fields generally could be identified by the presence of severe and irregular contractions and depressions that did not conform to the visual system's neuro-architecture. Further, functional fields generally presented one or more isolated threshold 'spikes', that is, isolated locations showing much better than average sensitivity. On repeated examinations, threshold spikes always changed locations. Visual evaluation for spikes proved superior to an objective computational algorithm. Fairly reliable objective discrimination of functional fields could be achieved by point-wise correlations of repeated examinations: median intertest correlation coefficients equalled 0.47 compared with 0.81 for true lesions. CONCLUSIONS Functional visual loss can be identified by automated static perimetry. Useful criteria include severe and irregular contractions and depressions, the presence of isolated threshold spikes and poor intertest correlations.