Many biomarkers for breast cancer prognosis have been proposed during the last two decades, among which HER2 and oestrogen receptors are of common use in routine clinical practice. However, in recent years, BCL2 has been recognized as an important prognostic parameter in human breast cancer, although its clinical utility is well established. The aim of this study was to examine the protein expression patterns of BCL2, HER2, oestrogen (ER) and progesterone receptors (PR) and to evaluate their correlation with survival and other prognostic parameters such as tumour size, histological grade and metastasis. We used a retrospective study including 84 Tunisian women with breast cancer. Immunohistochemistry was used to measure protein expression levels of several biomarkers. Association with conventional biopathological factors was analysed by SPSS (version13). The expression rates of BCL2, HER2, ER and PR were, respectively, 69, 62, 58.3 and 51.2%. In univariate analyses, BCL2 was highly correlated with both PR (P < 0.001) and ER (P = 0.006) and also with HER2 expression (P = 0.001). The triple negative profile showed a significant association with SBR (P = 0.016) and BCL2 expression (P = 0.02). In multivariate analyses, a significant association was maintained between BCL2 and both PR and ER (P = 0.02 and P = 0.004, respectively). Survival analysis showed that BCL2 expression was positively correlated with patients survival (P = 0.032). A Bayesian network analysis of all the variables confirmed the high value of BCL2 expression as a predictor of survival. As conclusion, BCL2 expression seems to be a very useful factor that should be in combination with HER2 and ER in breast cancer prognosis.