What is the incidence of aneuploidy and mosaicism in all cells of top-quality Day-4 embryos analysed by array-based comparative genomic hybridization (array CGH)?Our data show extensive abnormalities in Day-4 embryos.Numerous studies on human embryos at Day 3 and Day 5 of development show that they frequently contain aneuploid cells and are mosaic, although Day-5 embryos contain proportionally more normal cells than at Day 3. In contrast, only limited data exist on Day 4 of preimplantation development, despite the fact that it is the suggested stage for the initiation of the process of self-correction.Thirteen embryos were analysed: four fresh good-quality preimplantation genetic diagnosis (PGD) embryos and nine good-quality surplus embryos cryopreserved on Day 3 and donated for research. On Day 4, following removal of the zona pellucida, all blastomeres were disaggregated and collected.The genomic DNA of 283 single blastomeres from disaggregated embryos was amplified. Array CGH was carried out using 24SureTM Cytochip microarrays. After scanning of the microarray slides, the images were analysed using BlueFuse Software (BlueGnome). Combined with selective microsatellite analysis, hypothetical reconstructions of embryo chromosome complements were made following each of the first four cleavage divisions.No chromosome imbalance was detected for one PGD embryo, the other three were mosaic containing between 16 and 75% abnormal cells. All nine frozen-thawed embryos were abnormal. Six were mosaic with between 30 and 100% abnormal cells; three had abnormalities of meiotic origin, two of which displayed mitotic abnormalities. Evidence was also found of mitotic unbalanced structural chromosome rearrangements. The higher rate of abnormality of frozen-thawed embryos is based on a small number of embryos and cannot be tested statistically. The aneuploidy can mostly be explained by anaphase lag and non-disjunction. In some cases, we hypothesize endoreduplication followed by a cellular division with multipolar spindles to explain the results.Array CGH technology determines relative quantification of chromosomal domains but does not allow for the visualization of chromosomal rearrangements, assessment of ploidy or detection of uniparental isodisomy. Conclusions drawn on segmental abnormalities should be treated with caution. The division trees presented are hypothetical models projecting back in time that try to explain observations in single blastomeres of Day 4 embryos. The limited number of embryos analysed does not allow drawing firm conclusions, but nevertheless provides valuable data on the origin of aneuploidy in human embryos.Our data show extensive abnormalities in Day-4 embryos. We found no evidence of self-correction at this stage of development, suggesting that this process may start at a later stage of development.