The repair of DNA double-strand breaks by non-homologous end-joining (NHEJ) is efficient in non-dividing cells, but its use for site-specific transgene integration has not been shown. Suzuki et al. developed homology-independent targeted integration (HITI) for CRISPR–Cas9- and NHEJ-based gene knock-in. HITI successfully mediated site-directed GFP knock-in in postmitotic cultured mouse and human neurons. To improve applicability in vivo, HITI constructs were sub-cloned into adeno-associated virus (AAV). Following systemic delivery in mice, HITI was 90−95% on-target in muscle and heart tissues, with minimal NHEJ-associated mutagenesis. HITI–AAV was tested in the RCS rat, which is a model for retinitis pigmentosa (inherited retinal degeneration) caused by mutations in exon 2 of the Mertk gene. Injection of HITI–AAV−Mertk exon 2 into the eyes of RCS rats led to MERTK expression and this significantly improved retinal physiology and function.