Abstract Study question Is double-strand DNA break repair functional in human preimplantation embryos and what implications might a deficiency have for genome editing performed at early developmental stages? Summary answer CRISPR-Cas9 succeeds in generating DNA breaks with high efficiency, but these often remain unresolved, indicating that early human embryos are deficient in DNA repair. What is known already Genome editing (GE) technologies, such as CRISPR-Cas9, raise the possibility of avoiding transmission of inherited disorders by converting mutant genes back to wild-type (normal). The therapeutic application of CRISPR-Cas9 at the preimplantation stage is controversial due to uncertainty over safety and concerns about the alteration of the human germline. However, from a technical perspective it remains highly attractive, since it is the only developmental stage where delivery of CRISPR-Cas9 to every cell of the individual can be guaranteed. Currently, information on safety aspects and efficacy are lacking. Therefore, evaluation of the cellular response to CRISPR-Cas9 in human embryos is needed. Study design, size, duration 84 embryos were generated for research in an IRB approved study. For this purpose, donor oocytes were fertilised with donor sperm using ICSI. 51 of the resulting embryos served as controls, while in the other 33 double strand DNA breaks (DSBs) were created in a highly controlled fashion, directed at specific genomic sites using CRISPR-Cas9 technology. Successfully fertilised oocytes underwent culture in a time-lapse incubator and the duration of key developmental events were carefully timed. Participants/materials, setting, methods CRISPR-Cas9 components targeted several non-coding sites in the genome. An artificial DNA fragment was provided, which cells could potentially use for homology directed repair (HDR). Embryos were grown for 60 hours, then disaggregated. Blastomeres underwent whole genome amplification, followed by next generation sequencing to detect segmental aneuploidy related to failure of DNA repair. Additionally, sites targeted by CRISPR-Cas9 were amplified and sequenced to confirm whether repair had occurred and to distinguish the specific mechanism employed. Main results and the role of chance The study successfully validated and utilised three CRISPR ‘guides’ that targeted sites on chromosomes 2, 3 and 12. Alterations at the targeted sites were detected in 24/25 embryos (96% targeting efficiency). Repair of DSBs created by CRISPR-Cas9 was considered successful when an embryo contained at least one blastomere with changes to the DNA sequence at the targeted site consistent with nonhomologous end joining (NHEJ) or HDR. In total, 53 double-stranded breaks were generated by CRISPR-Cas9 during this study. Of these, 51% (27) were repaired by NHEJ, 9% (5) repaired by HDR and 40% (21) remained unresolved, the persistent breakage eventually causing segmental aneuploidy with a breakpoint at the targeted side. Segmental abnormalities are known to be detrimental to embryonic viability and risk congenital abnormalities in offspring. This study shows that human embryos, prior to activation of the genome, have a DNA damage repair deficiency. Timings of morphokinetic milestones were altered in embryos with unresolved DNA damage, confirming that checkpoint control is active. However, continued mitotic division, despite the presence of DSBs, shows cellular mechanisms that usually preserve genetic integrity lack stringency, ultimately failing to ensure repair. The results provide a warning against the therapeutic use of CRISPR-Cas9 in human embryos. Limitations, reasons for caution While targeting the embryonic genome using CRISPR-Cas9 was remarkably efficient (DSBs introduced in 96% of embryos), the majority of cells utilised NHEJ for repair, a process allowing introduction of additional mutations rather than correcting existing ones. Only <10% of cells used the HDR repair mechanism necessary for removal of mutations. Wider implications of the findings These findings reveal that cells of early human embryos are less able to process DSBs than other cell types. This suggests that embryo culture methods should give consideration to protecting the genome from damage. The results caution against clinical application of genome editing technologies that create DSBs, such as CRISPR-Cas9. Trial registration number NA