It has long been observed that many neuronal types position their nuclei within restricted cytoplasmic boundaries. A striking example is the apical localization of cone photoreceptors nuclei at the outer edge of the outer nuclear layer of mammalian retinas. Yet, little is known about how such nuclear spatial confinement is achieved and further maintained. Linkers of the Nucleoskeleton to the Cytoskeleton (LINC complexes) consist of evolutionary-conserved macromolecular assemblies that span the nuclear envelope to connect the nucleus with the peripheral cytoskeleton. Here, we applied a new transgenic strategy to disrupt LINC complexes either in cones or rods. In adult cones, we observed a drastic nuclear mislocalization on the basal side of the ONL that affected cone terminals overall architecture. We further provide evidence that this phenotype may stem from the inability of cone precursor nuclei to migrate towards the apical side of the outer nuclear layer during early postnatal retinal development. By contrast, disruption of LINC complexes within rod photoreceptors, whose nuclei are scattered across the outer nuclear layer, had no effect on the positioning of their nuclei thereby emphasizing differential requirements for LINC complexes by different neuronal types. We further show that Sun1, a component of LINC complexes, but not A-type lamins, which interact with LINC complexes at the nuclear envelope, participate in cone nuclei positioning. This study provides key mechanistic aspects underlying the well-known spatial confinement of cone nuclei as well as a new mouse model to evaluate the pathological relevance of nuclear mispositioning.