A TV monitor was used to evoke either a pattern ERG to a contrast-reversing checkerboard (PERG), or a focal ERG to alternate increases and decreases of luminance of the blank screen within a bright surround (FERG). Both responses are small (approx 2 microV) and fast (approx 50 msec to peak) and are similar in several other properties. However, they differ in timing and respond differently to changes in contrast. Each frame of a TV picture evokes a "raster ERG," even though the screen is blank. The response is focal and specific to a small central strip of the screen. It is simpler to record than the FERG, where the whole screen is flashing. Because the FERG summation area is about 4 deg, small squares (checks) reversing in contrast produce little luminance response. In 5 of 7 cases where the PERG is unilaterally reduced, the FERGs or raster responses were not affected. Thus clinical evidence also suggests that the PERG may be a separate phenomenon to the FERG and produced at a different site. Toxic, traumatic, congenital, and degenerative diseases of the optic nerve reduce the PERG. The comparison is most easily made in unilateral disease. Ten weeks after an optic nerve insult, the PERG becomes reduced in the affected eye as if retrograde degeneration was occurring. In 27 amblyopes of various types, the PERG was reduced in 23 where orthoptic treatment had failed. In 4 patients responding to treatment, PERGs of the amblyopic eye were as large as, or larger than, those of the fellow eye. The loss is greater with smaller checks. Retinal changes do occur after age 4 but so slowly that responses in heavily occluded eyes are not reduced. An additional level in the visual pathway is thus accessible to evoked potential investigation.