Visual evoked potentials (VEPs) were recorded from cat cortex (area 17) before, during and after application of the GABA blocker bicuculline (iontophoretic or topical). The stimuli comprised a test sinusoidal grating, and a mask grating oriented either parallel or orthogonal to the test. Both test and mask alternated in contrast at different temporal frequencies. VEPs were averaged in synchrony with the test contrast reversal, so the mask did not contribute directly to the averaged VEP response. Before application of bicuculline, both parallel and orthogonal masks attenuated the amplitude of VEPs and changed the phase response, but in different ways. Orthogonal masks lowered the slope of the contrast response curve without affecting extrapolated threshold, while parallel masks caused the curve to shift to the right. Orthogonal masks increased the phase advance, while parallel masks eliminated it. During application of bicuculline, neither parallel nor the orthogonal masks attenuated VEP amplitudes. The results suggest that although the mechanisms for the action of parallel and orthogonal masks are clearly distinct, both are mediated by the GABA-ergic inhibitory system. Given this evidence, measurement of VEP contrast response curves may provide a simple non-invasive technique for monitoring visual inhibition in humans.