The dicentric chromosome dic(9;20)(p11∼13;q11), occurs in 2% of the cases with pediatric acute lymphoblastic leukemia (ALL), almost exclusively of the B-cell lineage. Although previous cytogenetic studies have provided first insights into the genetic alterations in cases with dic(9;20), they were limited by the lack of sensitive genome-wide and next generation sequencing-based methods. Furthermore, the prognostic relevance of the dic(9;20) remains unclear, as previous studies have reported variable outcomes among these cases. Therefore, a better understanding of the (epi)genomic landscape and prognostic relevance of dic(9;20) ALL is essential to further optimize treatment of children with these alterations. The aim of this study was to unravel the demographic, clinical, prognostic and molecular characteristics of dic(9;20) in pediatric ALL. The present study investigates a cohort of 57 children with B-ALL, enrolled in the German ALL-BFM 2000, 2009, or 2017 trials, which underwent cytogenetic diagnostics in our center. Detection of dic(9;20) by conventional cytogenetic and molecular methods was the prerequisite for inclusion into the study. All 57 dic(9;20) cases, and 56 cases for which RNA samples were available, were subjected to genome-wide CGH+SNP array and targeted RNA-sequencing analysis. In addition, we performed whole transcriptome, whole genome and whole genome bisulfite sequencing for a selected number of cases. Event-free survival (EFS) was estimated according to Kaplan-Meier, and the curves were compared by log-rank test. Overall, the cohort consists of 22 male and 35 female children, with a median age at diagnosis of 3 years (range: 1-17 years). Patients were treated according to the AIEOP-BFM treatment protocols and stratified in the standard risk (n=15), medium risk (n=37) or high risk (n=5) treatment arms. All 57 patients achieved remission, 11 experienced relapses, while two patients in remission were lost during follow-up. Integrative analysis of RNA-seq and CGH+SNP array revealed recurrent genomic rearrangements of the DNMT3B gene with PAX5 or adjacent ZCCHC7 gene (n=3, each). DNMT3B is a de novo DNA methyltransferase and deletions of this gene have been previously reported in various solid malignancies and AML. The rearrangement of DNMT3B and ZCCHC7 resulted in an in-frame chimeric fusion transcript, containing the 5’ end of the open reading frame of DNMT3B and the 3’ end of ZCCHC7. In contrast, due to opposing orientation of the DNMT3B and PAX5 genes, the resulting chimeric transcript contained the 5’ end of the open reading frame of DNMT3B or PAX5 and a run-through transcript of the antisense strand of the partner gene. All genomic rearrangements with their breakpoints and fusion transcripts were validated using whole genome and Sanger sequencing. In the total group of six cases with rearrangements in the DNMT3B and PAX5/ZCCHC7 genes, four experienced relapses, resulting in a 5-year EFS of 25% (SE=20.4), compared to 75.4% (SE=8.2) for the remaining cohort (p=0.015). All six patients were treated according to the medium risk arm of the respective AIEOP-BFM 2000 or 2009 protocols (n=1 and n=5, respectively), and all showed good initial response to prednisone. Three patients had non-quantifiable levels of minimal residual disease after induction, while one of each had measurable levels ranging from 10-3-10-5. The unfavorable prognostic marker IKZF1plus was detected in four patients (66%), two of which relapsed. Since the DNMT3B gene functions as de novo methyltransferase, we examined whether genomic rearrangements in these cases can lead to global changes in methylation, but we did not observe either global hypermethylation, or global hypomethylation in dic(9;20) compared to cases of other B-ALL subtypes. In this study we describe novel recurrent genomic rearrangements involving the DNMT3B and PAX5/ZCCHC7 genes in pediatric ALL with dic(9;20). Although our observations should be confirmed in an independent cohort and larger series of patients, cases with rearrangements involving DNMT3B and PAX5/ZCCHC7 genes appear to have very poor outcome compared to other cases with dic(9;20). Once confirmed in an independent cohort, dic(9;20)-positive cases with rearrangements involving DNMT3B and PAX5/ZCCHC7 genes should be considered as high risk for relapse and treated accordingly.