To date, there is no effective marker to predict chemoresistance in cancers. In this study, we aimed to find a signature that can detect chemoresistance to taxane-based therapies in breast cancer. By studying the gene-expression profiling in discovery cohorts with 92 taxane-resistant and 68 taxane-sensitive patients, a 20-gene taxane-based chemotherapy signature (TAXSig) and a TAXSig equation were developed. The TAXSig and its equation were later validated in five further independent datasets with a total of 659 patients. In general, the TAXSig equation easily and effectively discriminated between chemoresistant and chemosensitive individuals. The TAXSig-identified groups showed significant differences in clinical outcomes both in estrogen-receptor-positive and -negative (ER(-)) breast cancer patients, while the TAXSig was especially powerful in identifying ER(-) patients who had a good prognosis and were chemosensitive. In conclusion, the TAXSig is a reliable, effective, and reproducible means of classifying chemoresistance to taxane-based therapies in breast cancer.