MicroRNAs (miRs) are small, non-protein coding transcripts involved in many cellular functions. Many miRs have emerged as important cancer biomarkers. In the present study, we investigated whether miR levels in breast tumors are predictive of breast cancer local recurrence (LR). Sixty-eight women who were diagnosed with breast cancer at the Lombardi Comprehensive Cancer Center were included in this study. Breast cancer patients with LR and those without LR were matched on year of surgery, age at diagnosis, and type of surgery. Candidate miRs were identified by screening the expression levels of 754 human miRs using miR arrays in 16 breast tumor samples from 8 cases with LR and 8 cases without LR. Eight candidate miRs that showed significant differences between tumors with and without LR were further verified in 52 tumor samples using real-time PCR. Higher expression of miR-9 was significantly associated with breast cancer LR in all cases as well as the subset of estrogen receptor (ER) positive cases (p = 0.02). The AUCs (Area Under Curve) of receiver operating characteristic (ROC) curves of miR-9 for all tumors and ER positive tumors are 0.68 (p = 0.02) and 0.69 (p = 0.02), respectively. In ER positive cases, Kaplan-Meier analysis showed that patients with lower miR-9 levels had significantly better 10-year LR-free survival (67.9% vs 30.8%, p = 0.02). Expression levels of miR-9 and another miR candidate, miR-375, were also strongly associated with ER status (p<0.001 for both). The potential of miR-9 as a biomarker for LR warrants further investigation with larger sample size.