We aimed in this study at identifying prognostic immunohistochemical molecular signatures indicative of disease outcome, also relevant for development of new specific therapies, in triple-negative (ER, PR, c-erbB2- negative) breast carcinoma subtypes. We evaluated 42 markers in tissue micro-arrays from a series of 924 breast carcinomas including 184 triple-negative tumors using standardized quantitative immunocytochemical assays and correlated the data with patients' outcome (mean follow-up of 79 months). When 27/42 markers including basal-like markers first found to be individually significant for prognosis in a univariate analysis (log-rank test) in 924 tumors, were secondly evaluated in the triple-negative tumor subtype (184/924), eleven including maspin, P21, P27, PTEN, caveolin, EGFR, FAK, P38, pMAPK, STAT1 and CD10 were 89.2% predictive of disease outcome in logistic regression. When markers reported in the literature as expressed in basal-like subtype were evaluated in the 924 series, only eight (EGFR, CK14, moesin, caveolin, cMet, ckit, CD44v6, C10) were prognosis predictive in univariate analysis (log-rank test) and in logistic regression were predictive of disease outcome in 66.3% independently of ER, PR and c-erbB2 expression and in 72% in triple-negative tumor subset. The results suggest that the category of 'triple-negative' breast carcinomas does not exactly overlap the basal-like subtype, and that immunoprofiling of triple-negative tumors (not similar to that of basal-like tumors) may be helpful to select patients for more aggressive treatment and provides a basis for development of tailored therapy.