Purpose: Novel noninvasive biomarkers with high sensitivity and specificity for the diagnosis of breast cancer (BC) are urgently needed in clinics. The aim of this study was to explore whether miRNAs from the miR-106a–363 cluster can be detected in the circulation of BC patients and whether these miRNAs can serve as potential diagnostic biomarkers. Methods: The expression of 12 miRNAs from the miR-106a–363 cluster was evaluated using qRT-PCR in 400 plasma samples (from 200 BC patients and 200 healthy controls (HCs)) and 406 serum samples (from 204 BC patients and 202 HCs) via a three-phase study. The identified miRNAs were further examined in tissues (32 paired breast tissues), plasma exosomes (from 32 BC patients and 32 HCs), and serum exosomes (from 32 BC patients and 32 HCs). Results: Upregulated levels of four plasma miRNAs (miR-106a-3p, miR-106a-5p, miR-20b-5p, and miR-92a-2-5p) and four serum miRNAs (miR-106a-5p, miR-19b-3p, miR-20b-5p, and miR-92a-3p) were identified and validated in BC. A plasma 4-miRNA panel and a serum 4-miRNA panel were constructed to discriminate BC patients from HCs. The areas under the receiver-operating characteristic curves of the plasma panel were 0.880, 0.902, and 0.858, and those of the serum panel were 0.910, 0.974, and 0.949 for the training, testing, and external validation phases, respectively. Two overlapping miRNAs (miR-106a-5p and miR-20b-5p) were consistently upregulated in BC tissues. Except for the expression of the plasma-derived exosomal miR-20b-5p, the expression patterns of exosomal miRNAs were concordant between plasma and serum, indicating the potential use of exosomal miRNAs as biomarkers. Conclusion: We identified four plasma miRNAs and four serum miRNAs from the miR-106a–363 cluster as promising novel biomarkers for the diagnosis of BC.