Human centromeres are composed of large tandem arrays of repetitive lpha-satellite DNA, which are often sites of chromosome rearrangements in cancers (Mitelman et al., 1997; Black & Giunta, 2018). Extensive sequence homology of the centromere tandem arrays suggests that homologous recombination plays an active role in maintaining the repeat structure, but also implies that this process must be tightly regulated to avoid rearrangements and loss of DNA, which could result in kinetochore malfunctions and cause genome instability. To date, a high fidelity linear sequence assembly is lacking for all except the Y chromosome, which has greatly hindered functional studies of these essential loci. Using the centromere chromosome-orientation fluorescence in situ hybridization (Cen-CO-FISH; Giunta, 2018) method, we showed that the centromere-specific histone variant CENP-A, and associated proteins CENP-C and CENP-T/W, are important for maintaining centromere integrity. Depletion of these proteins enhanced rearrangements at alpha-satellites independently of mitotic failure. This functionality is compromised in cancer cell lines and in primary cells undergoing senescence (Giunta & Funabiki, 2017). Although these results indicate the presence of active mechanism that prevents rearrangements at alpha-satellites, how centromere instability is induced under these circumstances is unknown. Using a rapidly inducible degradation system (Hoffmann et al., 2016), we have found that removal of CENP-A in S-phase leads to aberrant repeats recombination and delayed centromere replication, resulting in aberrant mitoses, progressive shortening of the centromere DNA and whole chromosome rearrangements in the subsequent cell cycle. I will present our latest finding that place human centromeres as a novel category of common fragile sites where replication burdens threaten repeats stability, with important implications for aneuploidy and carcinogenesis.