The Mr 52,000 cathepsin D is the precursor of a lysosomal protease secreted in excess by breast cancer cells. This protease can degrade extracellular matrices and proteoglycans and is induced by estrogens in estrogen receptor-positive breast cancer cell lines. In a 4- to 6-yr retrospective cohort study, the concentration of the total cathepsin D (precursor plus intermediate and mature chains) was assayed in cytosols of primary tumors from 242 pre/perimenopausal and 154 postmenopausal breast cancer patients in a solid-phase immunoassay using two specific monoclonal antibodies. Patients were initially divided into groups with low, intermediate, or high concentrations of cathepsin D corresponding to the quartiles of the overall distribution. Using these groupings, the level of Mr 52,000 cathepsin D was not significantly associated with the recognized prognostic factors of age, lymph node involvement, tumor size, and/or grade of anaplasia. A significant association was found between cathepsin D concentrations and estrogen receptor status only among pre/perimenopausal patients. Receptor-positive tumors (greater than or equal to 10 fmol of estrogen receptor/mg of cytosol protein) had a significantly greater proportion of patients with high Mr 52,000 cathepsin D concentrations. Patients with high Mr 52,000 cathepsin D concentrations (greater than 78 pmol/mg for pre/perimenopausal and greater than 24 for postmenopausal patients) have shorter recurrence-free survival (P = 0.06 for pre/peri- and P = 0.039 for postmenopausal patients) and have a trend toward shorter overall survival (P = 0.30 and P = 0.089 for pre/peri- and postmenopausal groups, respectively). In multivariate analysis, Mr 52,000 cathepsin D status was found to be an independent prognostic factor for recurrence-free survival of about the same import as lymph node status for both menopausal groups. This first retrospective study demonstrates that the level of Mr 52,000 cathepsin D in cytosol of primary breast cancer biopsies is an independent prognostic factor in predicting relapses in both pre/peri- and postmenopausal patients.