Key points The early receptor potential (ERP) of the mouse electroretinogram (ERG) was measured in wild‐type (WT) mice, in mice (Opn1sw−/−) that lack S‐cone opsin and overexpress M‐cone opsin, and in mice heterozygous for the retinal pigment epithelium isomerase Rpe65. The amplitude of the ERP saturated exponentially with flash intensity. In WT mice the saturated amplitude was ∼1000 μV, about 20% larger than the saturated amplitude of the ERG a‐wave. The ERP was 13% larger in mice overexpressing M‐opsin than in WT mice, indicating that 26% of the ERP of WT mice in these experiments arises from M‐opsin. After complete bleaching, the ERP of WT mice recovered in two phases, a fast phase responsible for ∼20% of the recovery having a time constant of ∼1 min, and a complementary slower phase with a time constant of 23 min. The fast phase of ERP recovery did not depend on the expression level of Rpe65, but the slow phase did. The fast phase of ERP recovery is concluded to arise from M‐opsin regeneration, and the slow phase from the regeneration of rod plasma membrane rhodopsin. The slower phase of ERP recovery is faster than the regeneration of bulk rhodopsin in the internal disc membranes, consistent with the hypothesis that delivery of 11‐cis retinal across the cytoplasmic gap between plasma and disc membranes retards regeneration of disc membrane rhodopsin. Abstract Sustained vertebrate vision requires that opsin chromophores isomerized by light to the all‐trans form be replaced with 11‐cis retinal to regenerate the visual pigment. We have characterized the early receptor potential (ERP), a component of the electroretinogram arising from photoisomerization‐induced charge displacements in plasma membrane visual pigment, and used it to measure pigment bleaching and regeneration in living mice. The mouse ERP was characterized by an outward ‘R2’ charge displacement with a time constant of 215 μs that discharged through a membrane with an apparent time constant of ∼0.6 ms. After complete bleaching of rhodopsin, the ERP recovered in two phases. The initial, faster phase had a time constant of ∼1 min, accounted for ∼20% of the total, and was not dependent on the level of expression of the retinal pigment epithelium isomerase, Rpe65. The slower, complementary phase had a time constant of 23 min in wild‐type (WT) mice (C57Bl/6) and was substantially slowed in Rpe65+/− mice. Comparison of the ERPs of a mouse line expressing 150% of the normal level of cone M‐opsin with those of WT mice revealed that M‐opsin contributed 26% of the total WT ERP in these experiments, with the remaining 74% arising from rhodopsin. Thus, the fast regenerating fraction (20%) corresponds approximately to the fraction of the total ERP independently estimated to arise from M‐opsin. Because both phases of the ERP recover substantially faster than previous measurements of bulk rhodopsin regeneration in living mice, we conclude that delivery of the highly hydrophobic 11‐cis retinal to the interior of rod photoreceptors appears to be retarded by transit across the cytoplasmic gap between plasma and disc membranes.