There is now an emerging body of evidence that shows there is a relationship between the survival of breast cancer patients and the expression level of steroid receptors. The aim of this study was to determine the relationship existing between estrogen receptors (ER) and progesterone receptors (PR) cytosolic content and the prognosis of postmenopausal breast cancer women under tamoxifen therapy. Two hundred and nineteen postmenopausal patients, without neoadjuvant chemotherapy and treated postoperatively with tamoxifen for at least 2 years, were followed up in our Cancer Center. We used flexible regression modeling and log likelihood methods for determining optimum cut-off values for steroid receptors, which allows the separation of patients into significantly different categories in term of survival. For PR, 3 categories were defined (category 1: PR < 10, category 2: 10 < or = PR < 60 and category 3: PR > or = 60 fmol/mg P). Univariate analysis at 8 years indicated that significant differences in event-free survival (EFS) were found for tumor size (T) (p = 0.005), lymph node status (N) (p = 0.003), histological Scarff, Bloom and Richardson grade (p = 0.003), ER values divided into 5 categories (p = 0.02) and PR values divided into 3 categories (p = 1 x 10(-5)). Eight-year EFS rate for the 3 PR categories, adjusted for N, were 39, 66 and 81%, respectively. Multivariate Cox analysis indicated that only T, N and PR values were significant variables for EFS. Patients with PR values > or = 60 present significantly greater EFS rates than patients with PR < 60 (p < 0.001). Our results show that the PR level in ER positive postmenopausal women is a strong prognostic marker in postmenopausal breast cancer women under tamoxifen therapy.