We studied the off-response of the rat ERG evoked with long duration, mesopic stimuli during light and dark adaptation, and after intravitreal injection of aspartate and (±)-cis-piperidine-2,3-dicarboxilic acid (PDA). At stimulus offset, the dark-adapted ERG always showed a rapid negative deflection followed by a positive deflection after which the potential returned to baseline. When the stimulus was turned off in the presence of a background of scotopic intensity, the positive deflection consisted of two components. One component was relatively small, fast, and insensitive to rod light adaptation. It resembled the d-wave of the rod ERG. The other component was slow and its amplitude grew with rod light adaptation. In the presence of aspartate, the fast-positive component was absent from the ERG while the remaining positive-going decay of the receptor potential had a time course similar to that of the slow-positive component in the untreated eye. Scotopically matched red and blue stimuli of mesopic intensity elicited equal ERG responses from the dark-adapted eye, including the two positive components in the off-response. These stimuli were also used to assess changes in the ERG off-response during recovery from a strong bleach. Even though the cone contribution to the rat ERG is very small, the presence of a small positive-going component in the off-response following an intense bleach suggested that this response originated from the cone pathway. PDA which suppresses the light response of hyperpolarizing bipolar cells and horizontal cells selectively eliminated the fast-positive component from the ERG. The findings of this study are inconsistent with the idea that the d-wave reflects the decay of the rod receptor potential. They support the possibility that signals from rods cross rod–cone gap junctions at mesopic light intensities, and drive second-order neurons in the cone pathway.