MicroRNAs (miRNAs) play a critical role on biological and cellular processes; the search for functional markers may be of importance for differential diagnosis, prognosis, and development of new therapeutic regimens. In this context, we evaluated the bone marrow miRNA profile of Brazilian children exhibiting T- or B-cell acute lymphoblastic leukemia (T-ALL or B-ALL), using massive parallel sequencing, using the HiSeq 2500 platform (Illumina). The differential expression analysis was conducted considering a leave-one-out approach and FDR ≤ 0.05. Machine learning algorithms were applied to search for the disease subset biomarkers. Target prediction, functional enrichment, and classification of biological categories were also performed. Sixteen miRNAs were differentially expressed between T- and B-ALL, of which 10 (miR-708-5p, miR-497-5p, miR-151a-5p, miR-151b, miR-371b-5p, miR-455-5p, miR-195-5p, miR-1266-5p, miR-574-5p, and miR-425-5p) were downregulated and six (miR-450b-5p, miR-450a-5p, miR-542-5p, miR-424-5p, miR-629-5p, and miR-29c-5p) were upregulated in childhood T-ALL. These miRNAs may be used for distinguishing childhood lymphoblastic leukemia subtypes, since it provided the clear separation of patients in these two distinct groups. Six relevant biological pathways were identified according to their role in leukemia, namely, viral carcinogenesis, cell cycle, and B-cell receptor signaling pathways for induced miRNAs and TGF-beta signaling, apoptosis, and NF-kappa B signaling for the repressed miRNAs, of which several miRNA gene targets participate in cell differentiation and hematopoiesis processes. Machine learning analysis pointed out miR-29c-5p expression as the best discriminator between childhood T- and B-ALL, which is involved in calcium signaling, critical for B-cell lymphocyte fate. Further studies are needed to assure the role of the 16 miRNAs and miR-29c-5p on acute lymphoblastic leukemia subtypes and on disease prognosis.