The adenosine triphosphate-based chemotherapy response assay (ATP-CRA) has the advantages of standardization, evaluability, reproducibility, and accuracy, and can be performed on relatively small numbers of tumor cells. A total of 43 patients were enrolled in the present study, and chemosensitivity tests were successfully performed in 40 (93.0%) of these patients. Twenty of the 40 received neoadjuvant chemotherapy or chemotherapy for metastatic breast cancer. The chemotherapy regimens used were doxorubicin plus docetaxel (n=9, 45.0%) or doxorubicin plus paclitaxel (n=11, 55.0%). Mean cell death rate, as determined by ATP-CRA, was lower in non-responders than in responders to therapy (P=0.012). Sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy for ATP-CRA were 78.6%, 100%, 100%, 66.7%, and 85.0%, respectively. Diagnostic accuracy achieved by immunohistochemistry using estrogen receptor or progesterone receptor was lower than that achieved using ATP-CRA. Expression of p53, erb-B2, Ki67, Bcl-2, Bcl-xL, and annexin I was not significantly associated with response to chemotherapy. Our results show that ATP-CRA has high specificity and positive predictive value for predicting response to chemotherapy.