Investigating neural stem cell (NSC) behaviour in vivo, which is a major area of research, requires NSC models to be developed. We carried out a multilevel characterisation of the zebrafish embryo peripheral midbrain layer (PML) and identified a unique vertebrate progenitor population. Located dorsally in the transparent embryo midbrain, these large slow-amplifying progenitors (SAPs) are accessible for long-term in vivo imaging. They form a neuroepithelial layer adjacent to the optic tectum, which has transitory fast-amplifying progenitors (FAPs) at its margin. The presence of these SAPs and FAPs in separate domains provided the opportunity to data mine the ZFIN expression pattern database for SAP markers, which are coexpressed in the retina. Most of them are involved in nucleotide synthesis, or encode nucleolar and ribosomal proteins. A mutant for the cad gene, which is strongly expressed in the PML, reveals severe midbrain defects with massive apoptosis and sustained proliferation. We discuss how fish midbrain and retina progenitors might derive from ancient sister cell types and have specific features that are not shared with other SAPs.