Purpose. To distinguish among three explanations of the in sensitivity to low frequency gratings, i. e., the low frequency cut in contrast sensitivity curves. Methods. We manipulated sensitivity to low frequencies by presenting gratings while simultanously increasing or decreasing mean luminance. The test gratings ranged from 0.25 to 18 c/deg, subtended 2.3, 4.5 or 9.1 deg, always had a mean luminance of 38 cd/m2, and were surrounded by a dark field. Contrast sensitivity was measured by forced choice between two 100 msec intervals separated by 500 msec. Between stimulus presentations the gratings were replaced by uniform fields of 0.1, 38, or 76 cd/m2. Both the contrast and luminance changed abruptly. Results. When the luminance was constant at 38 cd/m2 both between and during stimulus presentation, sensitivity to these abruptly presented gratings showed an attenuation at low frequencies, but the attenuation was not as great as is observed with gratings presented and removed gradually, as has been reported many times. However, when presentation of the grating was simultaneous with a change of mean luminance, from either 0.1 or 76 cd/m2 to 38 cd/m2, sensitivity to low spatial frequencies was profoundly reduced, but sensitivity to high spatial frequencies was hardly affected. The shapes of these curves cannot be accounted for by any combination of a spatially bandpass sustained channel and a spatially low-pass transient channel. Likewise, there is no reason that both increments and decrements of luminance should increase center-surround antagonism. However, the changes of mean luminance produce Fourier spectra that are of low spatial frequency and similar in the temporal domain to those of the gratings: hence, they mask test gratings of low spatial frequency. Conclusions. Neither conventional transient/sustained channels nor center-surround antagonism are sufficient to explain the low frequency cut, but an explanation based on implicit masking is, i. e., masking by the Fourier component at 0 c/deg,.