Background: There is currently a lack of biological markers to determine the risk of lymph node metastasis in breast cancer. A single long non-coding RNA (lncRNA) cannot accurately describe the heterogeneity of tumors. Thus, more accurate algorithms are needed to screen key pathogenic lncRNAs, and quantitative models are needed to describe the heterogeneity of breast cancer.Methods: A whole transcriptome sequencing data set of breast cancer tissue samples was downloaded from The Cancer Genome Atlas database (n=1,091). A weighted correlation network analysis was conducted to identify the hub lncRNAs associated with lymph node metastasis. A logistic regression analysis was conducted to construct the risk score model. The relationship between the risk scores and the key lncRNAs and the infiltration of the immune cell subtypes was also explored.Results: A total of 3 common lncRNAs were identified between the differentially expressed lncRNA set and the hub lncRNA set; that is, zinc finger protein 582-antisense RNA 1 (ZNF582-AS1), metastasisassociated lung adenocarcinoma transcript 1 (MALAT1), and actin filament associated protein 1-antisense RNA 1 (AFAP1-AS1). The following formula was used to calculate the risk score: risk score =1.31 + 0.51 * ZNF582-AS1 -0.66 * MALAT1 -0.50 * AFAP1-AS1. The receiver operating characteristic curve showed that the areas under the curve for the risk score, ZNF582-AS1, MALAT1, and AFAP1-AS1 were 0.975, 0.793, 0.685, and 0764, respectively (P<0.05). The risk score was positively correlated with immune cell subtype infiltration.Conclusions: ZNF582-AS1, MALAT1, and AFAP1-AS1 are the key lncRNAs involved in the lymph node metastasis of breast cancer. Our risk score model, which was based on ZNF582-AS1, MALAT1 and AFAP1-AS1, can accurately predict the risk of breast cancer lymph node metastasis. ZNF582-AS1, MALAT1, and AFAP1-AS1 are potential biomarkers for the lymph node metastasis of breast cancer.