In 83 elderly breast cancer patients, oestrogen and progesterone receptors (ER, PgR), pS2 and cathepsin D (CathD) were evaluated for their possible prognostic role on disease-free survival (DFS). The biomarkers were determined on the same cytosol by using immunoradiometric assays, and the variables were considered on a continuous scale. Univariate analysis indicated a linear relationship between logarithmic hazard ratio (log(HR)) and the log(ER) and log(PgR) concentration, but a non-linear relationship between log(HR) and CathD. As regards pS2, there was no evidence of a relationship with log(HR). In multivariate analysis, log(ER) content did not have a significant prognostic role, whereas log(PgR) retained a significant prognostic role. As regards the predictive ability, log(PgR) was the best discriminator of outcome followed by CathD, whereas the contribution of log(ER) was negligible. In multivariate analysis, 2 models were considered: one with log(ER), pS2, CathD and the interaction between pS2 and CathD, and another with log(PgR), pS2, CathD and the interaction between pS2 and CathD. In the first model, log(ER) content did not have a significant prognostic role, whereas in the second model log(PgR) retained a significant prognostic role. Our findings indicate that the quantitative determination of pS2 and CathD, in addition to steroid receptors, on the same cytosolic fraction could be a complementary tool to describe all breast cancer patients rather than just the elderly and that the use of a continuous scale, instead of a simple dichotomous "status", may improve the biological information supplied by the variables.