Psychophysical research on the Hermann grid illusion is reviewed and possible neurophysiological mechanisms are discussed. The illusion is most plausibly explained by lateral inhibition within the concentric receptive fields of retinal and/or geniculate ganglion cells, with contributions by the binocular orientation-specific cortical cells. Results may be summarized as follows: (a) For a strong Hermann grid illusion to be seen bar width must be matched to the mean size of receptive-field centers at any given retinal eccentricity. (b) With the use of this rationale, the diameter of foveal perceptive-field centers (the psychophysical correlate of receptive-field centers) has been found to be in the order of 4-5 min arc and that of total fields (centers plus surrounds) 18 min arc. These small diameters explain why the illusion tends to be absent in foveal vision. (c) With increasing distance from the fovea, perceptive-field centers increase to 1.7 deg at 15 deg eccentricity and then to 3.4 deg at 60 deg eccentricity. This doubling in diameter agrees with the change in size of retinal receptive-field centers in the monkey. (d) The Hermann grid illusion is diminished with dark adaptation. This finding is consistent with the reduction of the center-surround antagonism in retinal receptive fields. (e) The illusion is also weakened when the grid is presented diagonally, which suggests a contribution by the orientation-sensitive cells in the lateral geniculate nucleus and visual cortex. (f) Strong induction effects, similar to the bright and dark spots in the Hermann grid illusion, may be elicited by grids made of various shades of grey; and by grids varying only in chroma or hue. Not accounted for are: the illusory spots occurring in an outline grid ie with hollow squares, and the absence of an illusion when extra bars are added to the grid. Alternative explanations are discussed for the spurious lines connecting the illusory spots along the diagonals and the fuzzy dark bands traversing the rhombi in modified Hermann grids.