Scientific summary It has been known for decades that chromosomal trisomies and monosomies are a major cause of miscarriages. We hypothesized that smaller chromosomal imbalances might also lead to a significant number of pregnancy losses. With the introduction of higher resolution techniques, such as array CGH and next-generation sequencing, the detection of much smaller copy number changes and more complex rearrangements has become a possibility. This enabled us to test this hypothesis and to take a closer look at the origin and mechanism of segmental aberrations that occur during different stages of pre-conception, embryonic and fetal development. In the first part of this thesis, we used array CGH on miscarriage samples to search for submicroscopic CNVs. We initially investigated a series of spontaneous abortions by 1Mb resolution array CGH and conventional karyotyping. Array CGH showed a diagnostic advantage over conventional karyotyping. Aside from the ability to detect smaller chromosomal imbalances, the diagnostic yield was also improved through the direct extraction of DNA from the miscarriage tissues, because this avoids traditional problems such as culture failure, culture contamination and overgrowth with maternal cells. This was demonstrated by the detection of trisomies in cases without cell growth and by the identification of cases with low-grade mosaicism. As a consequence, 1Mb arrays were implemented as the standard technique for the cytogenetic analysis of miscarriages in our laboratory.In our search for the origin of early and recurrent miscarriages, we examined whether spontaneous abortions with a normal cytogenetic result after low resolution array CGH might carry copy number variations smaller than 1Mb that could account for the abortion. A genome wide screening for these small CNVs was applied on euploid embryos by means of high resolution SNP arrays. Analysis showed that more than half of the detected deletions and duplications were common CNVs also found in the general population and 87% were inherited from a parent. Taken together, this research demonstrated that, while CNVs <1Mb are present in miscarriages, the causality of these CNVs for the miscarriages themselves is very uncertain. SNP and B-allele frequency analysis of the euploid aborted embryos also showed that 4/11 embryos carried large (>5Mb) stretches of copy neutral LOH. Those data suggest some recurrent miscarriages may be due to the recurrent inheritance of pathogenic mutations from both parents that lead to recurrent lethal recessive disease. In the second part of this thesis, we studied the origin and mechanisms of chromosomal rearrangements often seen in miscarriage samples. First we focused on the origin of mosaic chromosomal abnormalities. It is generally assumed that they arise through postzygotic somatic mutations. To test this assumption, we determined the origin of mosaic segmental copy number changes in nine individuals. This revealed that in two out of nine cases the CNV originated during meiosis and thereby demonstrated that meiotic errors followed by two parallel postzygotic trisomy rescue events are a frequent cause of constitutional segmental mosaicism. In a second project we characterized twelve de novo unbalanced translocations by SNP typing and sequenced the breakpoints to the nucleotide level to determine when and how they arise. Detailed analysis of the translocations identified breakpoints in highly homologous transposable elements, predominantly in long interspersed elements (LINEs). This finding implies that non-allelic homologous recombination is the major driver behind de novo unbalanced translocations and reveals that a profoundly different mutational mechanism underlies unbalanced compared to balanced translocations. In addition, we detected several compound maternal/paternal derivative chromosomes, which reinforced the hypothesis that many chromosomal rearrangements arise postzygotically during early embryogenesis.