Recent findings have reported that human microRNAs (miRNAs) could serve as prognostic biomarkers in various cancers. We aimed to identify miRNAs that were associated with lymph node metastasis (LNM) and prognosis in breast cancer patients. A miRNA microarray covering 2019 mature miRNAs was used to identify differentially expressed miRNAs in 9 patients with LNM and 3 patients without LNM. Thirty-five differentially expressed miRNAs were identified, of which 10 significantly were up-regulated, whereas the other 25 were down-regulated in tissues with LNM compared with those without LNM. Seven miRNAs were subjected to quantitative real-time polymerase chain PCR (qRT-PCR) reaction, and 4 miRNAs (miR-191-5p, miR-214-3p, miR-451a, and miR-489) were validated in a total of 159 patients including a training set (n = 64) and a validation set (n = 95). The 4 miRNAs were used to construct a miRNA signature by logistic regression. Risk scores derived from the 4-miRNA signature were calculated to stratify the patients into high- or low-risk groups. Patients with high-risk scores had poorer overall survival and disease-free survival than did those with low-risk scores. The miRNA signature was an independent prognostic factor. MiR-191-5p increased, whereas miR-214-3p, miR-451a, and miR-489 inhibited cell proliferation, migration, and invasion abilities. The 4-miRNA signature may be a reliable prognostic and predictive tool for metastasis and survival in breast cancer patients.