Background: Childhood acute lymphoblastic leukemia (ALL) comprise a biologically heterogeneous group of malignancies, and its prognosis is strongly influenced by underlying genetic alterations. Although a comprehensive molecular classification has become essential for risk stratification and treatment, access to advanced genomic technologies remains limited in low- and middle-income countries. Here, we use RNA sequencing (RNA-seq) to characterize gene fusions and transcriptional subtypes in patients treated under the Brazilian public health system GBTLI protocol. Objectives: To establish a standardized workflow to identify gene fusions and gene expression–based subtypes of acute leukemia using RNA-seq data from Brazilian children enrolled at the GBTLI trial. Methods: Bone marrow or peripheral blood samples were collected at diagnosis from pediatric leukemia patients across all geographical regions of Brazil. RNA was extracted using TRIzol reagent and treated with RNase-free DNase. Libraries were prepared using the Illumina Stranded Total RNA Preparation assay and sequenced on the NovaSeq X Plus platform. Gene fusion detection was performed with the nf-core/rnafusion pipeline, considering true events when identified by at least two independent callers. Hyperdiploid B-ALL cases were identified using multiplex ligation-dependent probe amplification (MLPA), fluorescence in situ hybridization, and DNA index. RNA-seq data were processed with the nf-core/rnaseq pipeline for gene expression quantification. Ph-like ALL was assessed based on the expression of eight genes associated with the BCR::ABL1-like transcriptional signature. Results: Samples were obtained from 279 pediatric leukemia patients, including B-ALL (n=234), and T-ALL (n=45). There was a predominance of male patients (56% and 73%) and non-white individuals (53 and 71%). Most patients presented with leukocyte counts ≤50×10?/L and high blast percentages (>30%). To date, 96 samples have undergone RNA-seq analysis. In B-ALL, ETV6::RUNX1 was the most frequent subtype (27%), followed by B-other (24%) and hyperdiploidy (12%). Rare high-risk fusions, including TCF3::HLF, PAX5-r, ABL1-r, and CRLF2-r were identified. Hierarchical clustering revealed four cases with a Ph-like signature. In T-ALL, STIL::TAL1 and ABL1-r were the most recurrent fusions (23% each), along with NUP98-r, NOTCH1-r, and KMT2A-r. Importantly, 64% of previously unclassified cases were reassigned to defined molecular subgroups. In addition, we identified rare fusions, including KMT2A::USP2, MEF2D::BCL9, and CREBBP::SLX4. Regarding secondary alterations in ALL, CDKN2A/B deletions were the most frequent ones (31% and 64%). In B-ALL, deletions of ETV6 (21%) and IKZF1 (16%) genes were also recurrently observed. Copy-number alterations were associated with B-ALL subtypes. For instance, the BCR::ABL1 subtype had an enrichment of CDKN2A/B (55%) and IKZF1 (65%) deletions, while the ETV6::RUNX1 subtype often presented ETV6 deletions (56%). Conclusions/Implications: This study demonstrates the feasibility and clinical relevance of RNA-seq–based molecular profiling for childhood leukemia in a large, heterogeneous Brazilian cohort. The approach enabled the identification of recurrent and rare genetic subtypes, thus improving molecular classification, and uncovering high-risk lesions with potential therapeutic implications. These findings contribute to a more accurate understanding of the molecular alterations of pediatric leukemia in Brazil, thus supporting appropriate treatment under GBTLI Brazilian protocol.