We have recently described the urokinase-type plasminogen activator (uPA) and its type 1 inhibitor (PAI-1) as strong prognostic variables in breast cancer (J. A. Foekens et al., Cancer Res., 52: 6101-6105, 1992; J. Grondahl-Hansen et al., Cancer Res., 53: 2513-2521, 1993; J. A. Foekens et al., J. Clin. Oncol., 11: 899-908, 1994). A specific cell surface receptor (uPAR) binds uPA and strongly enhances plasmin generation, and the amount of uPAR in the tumor tissue might therefore be a rate-limiting factor in the extracellular proteolysis involved in tumor invasion. Here, we report on the prognostic value of uPAR in cytosolic (uPARc) and Triton (uPARt) extracts prepared from 505 primary breast tumors. The median observation time was 54 (range: 12-125) months. uPAR levels were determined by a sandwich ELISA. Univariate analysis showed that high uPAR levels (above the median value) were significantly associated with a shorter overall survival, showing a stronger discriminatory effect for uPARc [relative hazard rate (RHR): 1.47; P = 0.012)] as compared with uPARt (RHR, 1.33; P = 0.059), while no statistically significant differences were found for relapse-free survival. Multivariate analysis including all patients showed that when including other biochemical variables (estrogen receptor, progesterone receptor, PS2, cathepsin D, uPA, and PAI-1), the only retained independent variable via backward elimination was PAI-1 for both relapse-free survival and overall survival. When analyzed separately in clinically relevant subgroups, the prognostic value of uPAR was particularly strong in a subgroup of 201 node-positive postmenopausal women, showing considerably shorter overall (RHR: 2.39; P < 0.0001) and relapse free (RHR: 1.91; P = 0.0006) survival for patients with high uPARc content. High uPARt levels were also significantly associated with shorter overall survival in this subgroup of patients (RHR: 1.5; P = 0.047), but not with relapse-free survival (P = 0.64). Multivariate analysis, including the basic model, estrogen and progesterone receptors, PS2, cathepsin D, uPA, PAI-1, uPARc, and uPARt in the subgroup of postmenopausal node-positive patients, showed that only uPARc and PAI-1 were significant independent prognostic parameters, with respect to overall survival, RHRs being 2.72 (P < 0.0001) and 1.81 (P = 0.005), respectively. In multivariate analysis of relapse-free survival, uPARc, PAI-1, and uPA were independent parameters with respective relative relapse rates of 1.91 (P = 0.002) for uPARc, 1.68 (P = 0.02) for PAI-1, and 1.6 (P = 0.03) for uPA. These data lend support to the hypothesis that uPAR is an important molecule in plasmin-mediated extracellular matrix degradation leading to cancer cell dissemination and death of the patient.