The factors determining the clinical relevance of disseminated tumor cells (DTC) in breast cancer patients are largely unknown. Here we compared the specificity and clinical performance of two antibodies frequently used for DTC detection. Reactivities of antibodies A45-B/B3 (A45) and AE1/AE3 (AE) for selected cytokeratins (CK) were assessed by 2-DE Western Blot analysis. Using these antibodies bone marrow aspirates from 391 breast cancer patients (M(0), pT1-3, pN0-3) were screened for the presence of DTC. To obtain prognostic information, patients were followed up over a median of 83 months for time to relapse and 99 months for time to death. Among the analyzed CK, AE detected CK5, CK7, CK8, and CK19, whereas A45 recognized CK7 and CK18. In total, 24 of 391 patients (6.1%) were DTC-positive for A45, and 41 (10.5%) for AE. Although concordance between the two antibodies was 84.4%, overlap among positive cases was only 3.2%. DTC-positivity with AE and A45 was more frequent in patients of higher nodal status (P=0.019 and P=0.036, respectively). Nearly all patients with A45-positive DTC had hormone receptor-positive tumors (23/24), while detection of AE-positive DTC was more frequent among hormone receptor negative patients (P=0.006). Survival analyses of all patients revealed shorter distant disease-free survival (P=0.039) for patients with A45-positive DTC, whereas the prognostic relevance of AE-positive DTC was restricted to node-positive patients. The clinical utility of immunocytochemical (ICC) DTC detection depends on the anti-CK antibody used, which may reflect the complex CK composition of DTC.