The alterations of glycosylation, which is a common post-translational modification of proteins, have been acknowledged as key events in breast cancer (BC) oncogenesis and progression. The aberrant expression of glycosyltransferases leads to aberrant glycosylation patterns, posing the diagnostic potential in BC outcomes. The present study aims to establish a glycosyltransferase-based signature to predict BC prognosis and response to immune checkpoint inhibitors. We firstly screened 9 glycosyltransferase genes from The Cancer Genome Atlas (TCGA) database and accordingly established a glyco-signature for predicting the prognosis in BC patients. Patients with BC were successfully divided into high-risk and low-risk groups based on the median cutoff point for risk scores in this signature. Next, the combinational analyses of univariate and multivariate Cox regression, Kaplan-Meier, and receiver operating characteristic (ROC) curves were used to prove that this glyco-signature possessed excellent predictive performance for prognosis of BC patients, as the high-risk group possessed worse outcomes, in comparison to the low-risk group. Additionally, the Gene Set Enrichment Analysis (GSEA) and immunologic infiltration analysis were adopted and indicated that there was a more immunosuppressive state in the high-risk group than that in the low-risk group. The clinical sample validation verified that glycosyltransferase genes were differentially expressed in patients in the low- and high-risk groups, while the biomarkers of antitumor M1 macrophages were increased and N-glycosyltransferase STT3A decreased in the low-risk group. The final in vitro assay showed that the silencing of STT3A suppressed the proliferation and migration of BC cells. Collectively, our well-constructed glyco-signature is able to distinguish the high- and low-risk groups and accordingly predict BC prognosis, which will synergistically promote the prognosis evaluation and provide new immunotherapeutic targets for combating BC.