Growing evidence indicates that microRNAs (miRNAs) play critical roles in the initiation and progression of breast carcinoma (BC) and are promising diagnostic biomarkers. In the present study, we aimed to identify a multi-marker miRNA pool with high diagnostic performance for BC. We collected miRNA expression profiles of BC samples and normal breast tissues from The Cancer Genome Atlas (TCGA) and screened differentially expressed miRNAs by conducting two‑sample t-tests and by calculating log2 fold-change (log2FC) ratios. Statistical significance was established at p<0.001 and |log2FC| >1. Then, we generated receiver operating characteristic (ROC) curves, calculated the area under the curve (AUC) with a 95% confidence interval (95% CI), and calculated the diagnostic sensitivity and specificity using MedCalc software. Additionally, we predicted the targets of candidate miRNAs using 10 online databases: TarBase, miRTarBase, TargetScan, TargetMiner, microRNA.org, RNA22, PicTar-vert, miRDB, PITA and PolymiRTS. Target genes that were predicted by at least four algorithms were chosen, and cooperative targets of multiple miRNAs were further selected for GO and KEGG pathway analyses through the DAVID online tool. Eventually, a total of 66 differentially expressed miRNAs were identified after miRNA expression profiles were analyzed in BC and normal breast samples. Of these, we selected nine dysregulated miRNAs as candidate diagnostic markers: seven upregulated miRNAs (hsa-miR-21, hsa-miR-96, hsa-miR-183, hsa-miR‑182, hsa-miR-141, hsa-miR-200a and hsa-miR-429) and two downregulated miRNAs (hsa-miR-139 and hsa-miR‑145). The ROC curve for the combination of these nine differently expressed miRNAs showed extremely high diagnostic accuracy, with an AUC of 0.995 (95% CI, 0.988‑0.999) and diagnostic sensitivity and specificity of 98.7 and 98.9%, respectively. In conclusion, the combination of these nine miRNAs significantly improved the accuracy of breast cancer diagnosis.