INTRODUCTION: The addition of taxanes (Ts) to chemotherapeutic regimens has not demonstrated a consistent benefit in early-stage breast cancer. To date, no clinically relevant biomarkers that predict T response have been identified. METHODS: A dataset of immunohistochemistry stains in 411 patients was mined to identify potential markers of response. TLE3 emerged as a candidate marker for T response. To test the association with T sensitivity, an independent 'triple-negative' (TN) validation cohort was stained with anti-TLE3 antibody. RESULTS: TLE3 staining was associated with improved 5-year disease-free interval (DFI) in the overall cohort (n = 441, P < 0.004), in patients treated with cyclophosphamide (C), methotrexate, and 5-fluorouracil (n = 72, P < 0.02), and in those treated with regimens containing doxorubicin (A) and a T (n = 65, P < 0.04). However, no association was shown with outcome in untreated patients (n = 203, P = 0.49) or those treated with a regimen containing A only (n = 66, P = 0.97). In the TN cohort, TLE3 staining was significantly associated with improved 5-year DFI in all patients (n = 81, P < 0.015), in patients treated with AC + T (n = 45, P < 0.02), but not in patients treated with AC (n = 17, P = 0.81). TLE3 was independent of tumor size, nodal status, and grade by bivariable analysis in both cohorts. CONCLUSION: TLE3 staining is associated with improved DFI in T-treated patients in two independent cohorts. Since the validation study was performed in a TN cohort, TLE3 is not serving as a surrogate for estrogen receptor or HER2 expression. TLE3 should be studied in large clinical trial cohorts to establish its role in T chemotherapy selection.