OBJECTIVE We investigated whether basal cytokeratin (CK5/6 or CK17) expression had an impact on survival in patients with operable breast cancer. METHODS Expression of CK5/6 or CK17 was analyzed by immunohistochemistry in 195 women with breast cancer. RESULTS In total, 72 (37%) tumor samples were regarded as being positive for CK5/6 or CK17. The basal-like phenotype as defined by basal cytokeratin expression, lack of estrogen receptor (ER) and absence of HER2 overexpression was found in 48 (25%) cases. Positive staining for CK5/6 or CK17 was associated with worse prognosis when compared with patients negative for basal cytokeratins in all cases (5-year cancer-specific survival rate 59.4 vs. 77.5%, p = 0.0273) and in the node-negative group (70.5 vs. 90.8%, p = 0.0208) but not in the node-positive group (43.9 vs. 65.4%, p = 0.1182). To determine the real prognostic value of basal cytokeratins, survival in a group of ER-negative patients was analyzed depending on CK5/6 or CK 17 expression. No influence on survival was observed. The outcome of patients whose cancers were positive for cyclin E regardless of ER status was not changed by CK5/6 or CK17 expression. In multivariate analysis, independent prognostic factors affecting survival in the whole group included: nodal involvement, HER2 status and cyclin E expression. Neither ER status nor basal cytokeratin expression retained statistical significance. CONCLUSIONS We demonstrated that the poor prognosis associated with the basal-like phenotype of breast cancer was determined by ER absence and cyclin E expression and not by CK5/6 or CK17 expression.