Triple-negative breast cancer (TNBC) represents 10–20 % of all breast cancer (BC) cases and is characterized by poor prognosis. Given the urgent need to improve prognostication and develop specific therapies for TNBC, the identification of new molecular targets is of great importance. MicroRNA (miRNA) has been reported as a valuable and novel molecular target in the progression of TNBC. However, the expression and function of miRNAs in different tumors are heterogeneous. Herein, we first analyzed miRNA data from The Cancer Genome Atlas (TCGA) and surprisedly found that overexpressed miRNAs were associated with poor survival in all breast cancer patients, but the overexpressed miRNAs were associated with better survival in TNBC patients. Based on the heterogeneity of miRNA expression in TNBC, we conducted further analysis using univariate Cox proportional hazard regression models and identified 17 miRNAs with prognostic potential. Subsequently, a multivariate Cox model was employed to create a 3-miRNA prognostic model for predicting overall survival in TNBC patients. The diagnostic model exhibited an area under the curve (AUC) of 0.727, and multivariable Cox regression indicated that each covariate was associated with survival. These data indicate that this model is relatively accurate and robust for risk assessment, which have a certain value for clinical application. In order to explore the network behind the overexpressed miRNAs in TNBC, we established a novel network consisting of lncRNAs, miRNAs, and mRNAs through complete transcriptome data from matched samples in the TCGA database. In this network, IRS-1 appeared to be the top hub gene. Experimental results demonstrated that miR-15b-5p and miR-148a-3p effectively target IRS-1 in vitro, shedding light on the intricate regulatory mechanisms in TNBC mediated by the heterogeneous miRNAs. Besides, miR-148a-3p significantly inhibited cell migration and viability. Overall, this study may add valuable insights into the molecular landscape of TNBC based on miRNAs and have the potential to contribute to the development of targeted therapies and improved prognostic strategies of TNBC.