Background: Advanced breast cancer commonly metastasises to the bone and the molecular mechanism explaining the bone affinity of breast cancer cells is unclear. Thus, we developed nomograms based on a competing endogenous RNA (ceRNA) network and analysed tumour-infiltrating immunecells to elucidate the molecular pathways that may predict the prognosis of breast cancer patients.Methods: We obtained the RNA expression profile of 1091 primary breast cancer samples from The Cancer Genome Atlas database, 58 of which had bone metastasis. We analysed differential RNA expression patterns between breast cancer with and without bone metastasis and developed a ceRNA network. Cibersort was employed to differentiate between immune cell types based on tumour transcripts. Nomograms were then established using the ceRNA network and immune cell analysis. The value of prognostic factors was evaluated by Kaplan-Meier survival analysis and Cox proportional risk model.Results: There were significant differences in lncRNAs, 18 miRNAs, and 20mRNAs between breast cancer with and without bone metastasis, which were used to construct a ceRNA network. We found that the protein-coding genes gjb3, cammv, ptprz1,and fbn3 were significant in our Kaplan-Meier analysis. We also observed significant differences in plasma cell and follicular helper T cell populations between the two groups. In addition, the proportions of mast cells, gamma delta T cells, and plasma cells differed depending on disease location and stage. Our analysis revealed that a high proportion of follicular helper T cells and a low proportion of eosinophils promoted survival and that dlx6-as1, wnt6,and gabbr2expression may be related to bone metastasis of breast cancer.Conclusions: We provided a bioinformatic basis for exploring the molecular mechanism of bone metastasis in breast cancer patients and identified factors that may predict this.