BACKGROUND: The efficacy of high-dose chemotherapy with progenitor-cell rescue for women with breast cancer is a controversial issue. Although historically controlled trials have suggested a survival advantage for high-dose chemotherapy, several randomised studies have yet to confirm this advantage. Two studies, however, by Bezwoda, of patients with high-risk and metastatic disease, seemed to show a significant survival advantage for high-dose compared with conventional-dose chemotherapy for metastatic and high-risk primary breast cancer. METHODS: To corroborate the study results before starting a large international confirmatory study, a US team did an on-site review of records for patients in the high-risk study. Limited numbers of records were made available for review, all of which were for patients who received the high-dose-chemotherapy regimen. FINDINGS: There was much disparity between the reviewed records and the data presented at two international meetings. In addition, the reviewers saw no signed informed consent, and the institutional review committee had no record of approval for the investigational therapy. After the site visit, Bezwoda admitted scientific misconduct by using a different control chemotherapy regimen from that described in presented data. INTERPRETATION: The Bezwoda study should not be used as the basis for further trials to test the efficacy of the cyclophosphamide, mitoxantrone, etoposide regimen for high-dose chemotherapy in women with high-risk primary breast cancer. This review validates the essential nature of on-site audits, especially in single-institution studies.