Spectral-Domain Optical Coherence Tomography (SD-OCT) is a widely used method to observe retinal layers and follow pathological events in human. Recently, this technique has been adapted for animal imaging. This non-invasive technology brings a cross-sectional visualization of the retina, which permits to observe precisely each layer. There is a clear expansion of the use of this imaging modality in rodents, thus, a precise characterization of the different outer retinal layers observed by SD-OCT is now necessary to make the most of this technology. The identification of the inner strata until the outer nuclear layer has already been clearly established, while the attribution of the layers observed by SD-OCT to the structures corresponding to photoreceptors segments and retinal pigment epithelium is much more questionable. To progress in the understanding of experimental SD-OCT imaging, we developed a method for averaging SD-OCT data to generate a mean image allowing to better delineate layers in the retina of pigmented and albino strains of mice and rats. It allowed us to locate precisely the interface between photoreceptors and retinal pigment epithelium and to identify unambiguously four layers corresponding to the inner and outer parts of photoreceptors segments. We show that the thickness of the various layers can be measured as accurately in vivo on SD-OCT images, than post-mortem by a morphometric analysis of histological sections. We applied SD-OCT to different models and demonstrated that it allows analysis of focal or diffuse retinal pathological processes such as mutation-dependent damages or light-driven modification of photoreceptors. Moreover, we report a new method of combined use of SD-OCT and integration to quantify laser-induced choroidal neovascularization. In conclusion, we clearly demonstrated that SD-OCT represents a valuable tool for imaging the rodent retina that is at least as accurate as histology, non-invasive and allows longitudinal follow-up of the same animal.