Hepatoblastoma (HB) is a very rare embryonic tumor, although it is the most common liver tumor in children. HBs genomes contain few genetic alterations, being mainly characterized by recurrent aneuploidy gains of whole chromosome and / or whole chromosomal arms; However, the contribution of these changes to the genesis of HB is still poorly understood. Segmental genomic copy number alterations (CNAs) have also been described in these tumors, restricted to specific regions of some chromosomes, some of them in association with clinical features such as progression and poor prognosis. The main aim of this study was characterizing the CNA pattern of 25 HBs by cytogenomic analysis using genomic microarrays (chromosomal microarray analysis - CMA), considered the gold standard methodology for detection of submicroscopic CNAs (<4 Mb). Based on the results obtained by CMA, the 25 tumors could be classified in three categories according to their genomic profile: seven HBs showed no detectable alteration, thirteen presented few CNAs (up to four chromosomal alterations and/or focal alterations) and five tumors carried many CNAs, with complex genomes. One hundred and twenty-three CNAs were detected in the tumor group, and the most prevalent CNAs in the HBs group were gains at 1q, 2 / 2q and 20 / 20p, and losses at 1p / 1pter and 4 / 4q. In addition, it was possible to delimitate six chromosomal regions most commonly affected by CNAs in the cohort (minimum common regions - MCRs), located at: 1q31.3q42.3, 2q24.2q24.3, 20p13p11.1 (gains) and 1p36.31, 4p14, 4q21.22q25 (losses). In silico evaluation of the regions indicated 17 genes ((CHD5, BLACAT1, KDM5B, PARP1, LINC00210, MDM4, CHI3L1, PCAT6, PROX1, UBE2T, DPP4, FAP, HPSE, LEF1, NFKB1, PTPN13 and JAG1) as potentially related to tumorigenesis and/or HB progression, being candidates for future gene expression studies. Among these genes, only two were previously related to HB, both mapped to the MCR 1q31.3q42.3 gain: MDM4 and PARP1. In the cohort, 11 tumors presented this MCR 1q31.3q42.3 gain, and considering clinical data, ten of them were either high risk (6) or intermediate risk (4), with four deceased patients. Given the number of high/intermediate risk patients in this subgroup of tumors, 1q31.3q42.3 gain may be associated with more aggressive tumors. The second aim of this work was comparing the performance of three different next generation sequencing platforms (NGS), whole exome sequencing, OneSeq panel and TruSight One panel, with the gold-standard technique for detection of CNAs (CMA). Our results, based on data from 19 HBs, showed high rate of concordance between CMA and NGS data (~87%); the vast majority of the discordant cases (CNAs detected only by aCGH or NGS) were regions without coverage in one of the platforms. The minimum size of CNA detected in the NGS data was 109 Kb, not detected by CMA due to lack of coverage. We also found that it is possible to detect mosaic CNAs in NGS data (minimum mosaic frequency of ~10% for losses and ~20% for gains), as well as homozygous regions or cnLOH (copy-neutral losses of heterozygosity). However, there are limitations using NGS data based on capture (exome and panels), which can impair the detection of the full CNA spectrum and precise size mapping. Moreover, it was found that sequencing depth is a major issue for CNA detection using NGS data, since regions with low depth coverage have impaired detection or size delimitation. Despite these limitations, the great potential for obtaining accurate results both for mutations and CNAs allows the replacement of CMA by NGS platforms as the first line of genomic research