The amplitude of the leading edge of the a-wave of the human electroretinogram (ERG) was compared with predictions from a computational model of the light-induced responses of rod mammalian receptors. According to this model, a linear process describes the amplitude and time course of the response to relatively low flash intensities and at brief times after the onset of the flash. At higher flash intensities, a nonlinear process, described by the Naka-Rushton function or a saturating exponential, is involved. The primary focus here is on intensity-response data recorded with a clinical ganzfeld apparatus. The leading edge of the rod a-wave recorded from normal observers and patients with congenital stationary night blindness (CSNB) was described by a linear process for flash intensities up to the maximum available flash intensity, 2.0 log scot td-sec. This finding is consistent with the model of the rod's response. It suggests, however, that when ERGs are recorded with clinical systems limited to 2.0 log scot td-sec, these data cannot be used to distinguish between changes in the parameters (eg, semisaturation intensity versus maximum response) of the human rod receptors. Responses to flash intensities up to 3.4 log scot td-sec were recorded using a custom, high-intensity ganzfeld system. Both the linear and nonlinear components of the model were needed to fit the ERGs recorded with this system. This suggests that changes in different receptor parameters can be distinguished with higher intensity flashes.