Goals: To determine differences in TNBC between expression of cytokeratins 5 and 6 (CK 5/6) and their prognosis. Methods: Data from 888 consecutive female patients with invasive breast cancer but without metastatic disease, who had undergone surgery and were registered between 1 January 1997 and 31 December 2004 in the Tumour Registry of the University Hospital of Ourense (CHUO), Galicia, Spain, were collated and analysed. 100 of them were triple negative breast cancer. IHC was then performed on archived biopsies from the patients, who were thus classified by breast cancer subtype. First, a descriptive analysis of the study population was performed. Qualitative variables were analysed as absolute percentages and relative frequencies, and quantitative variables in terms of central tendency values, i.e. mean. Statistical analyses were then performed, using statistical tests appropriate to each type of variable, i.e. Chi-square test or Fisher’s exact test. Global survival and disease-free survival, using Cox regressions and Kaplan–Meier estimates, were calculated for classical prognostic variables (included age, tumour size and grade, node involvement). We present the results of triple negative breast cancer cases, which were divided in “basal like” and “no basal like” by CK 5/6 expression. SPSS version 17.0 (SPSS Inc., Chicago, IL) was used for all statistical analyses. Results were considered significant when p < 0.05. Results: One hundred triple negative (11.26%) cases among a total of 888 breast cancers were identified: 48 basal like (48%), 45 (45%) no basal like and 7 of them unclassified, because it was not possible to find histological samples. The basal like subtype showed CK 5/6 expression and no basal like absence of this expression. The relapse happened in 28% of TNBC. 42% of the relapses were visceral metastasis, 10% in nervous central system (CNS) and 7% in bones. There was more recurrences in CNS in basal-like than no basal like subtype (11 versus 7%) There was much more mortality-related breast cancer in basal like than in patients with the no basal like subtype (23% versus 18%) and much higher non-cancer-related mortality in patients with no basal like subtype. Patients without node involvement (N0) and basal like subtype had more relapses than no basal like subtype (22 versus 12.5%). None of the no basal like subtype patients died. Nevertheless, 22% of patients of basal like subtype died. Conclusion: TNBC is a bad prognostic factor in multivariate analyses, independent of size, nodal involvement and grading. By immunohistochemical expression of CK 5/6 we classified TNBC in basal like and no basal like subtypes, founded prognostic differences. Even though patients with no basal like tumors relapse, it seems that they have a lesser aggressive behavior. In contrast, basal like subtype tumors have a threatening prognosis with a short time between relapse and death. This evidence must make us think about the consideration of more intense adjuvant therapy in basal like TNBC to decrease relapse and death. Disclosure of Interest: No significant relationships.