The present study aimed to investigate the clinical relevance of disseminated tumor cells (DTC) in breast cancer patients before and after high-dose adjuvant chemotherapy with or without progenitor stem-cell support. One hundred and eighteen high-risk stage II breast cancer patients entering the Scandinavian Study Group multicenter trial were randomized to 9 cycles of tailored and dose-escalated FEC (5-fluorouracil, epirubicin, cyclophosphamide) or 3 cycles of standard FEC followed by high-dose chemotherapy. Bone marrow (BM) samples at diagnosis and 6 months after completion of chemotherapy were assessed for the presence of cytokeratin positive (CK+) cells. Before treatment, 29% of the patients were CK+ (21% in the dose-escalated group and 36% in the high-dose-group). Six months after treatment, 17% of the patients were CK+ (17 and 16% respectively). Of the 95 patients who were evaluated 6 months after treatment, 60% were consistently CK-. CK+ cells in BM was evaluated as a prognostic and predictive marker and compared to other defined prognostic factors of the primary tumor. Monitoring BM changes at the time of diagnosis and 6 months posttreatment is an independent predictive factor for breast-cancer-specific survival (BCS) (p = 0.001). Those who have consistent CK negative (-) BM findings constitute a group of patients with good prognosis. Our results suggest that changes in CK+ cells in BM before and after chemotherapy can be used clinically as a surrogate maker to predict outcome in breast cancer patients.