Evaluation of the relationship between plasminogen activator inhibitor-1 (PAI-1) and the metastatic potential of primary breast cancer, and to compare the prognostic impact of PAI-1 in multivariate analysis with those of conventional prognostic factors, including steroid-hormone receptors, and those of urokinase plasminogen activator (uPA), pS2-protein (PS2), and cathepsin D. Cell biologic prognostic factors were analyzed in 657 cytosols routinely prepared from frozen-tissue biopsies that were submitted to our laboratory for the assessment of steroid-hormone receptor status. The median duration of follow-up in patients still alive at the time of analysis was 48 months. Estrogen receptor (ER) and progesterone receptor (PgR) status were assessed by radioligand binding assay, PS2, and cathepsin D by radiometric immunoassay, and uPA and PAI-1 by enzyme-linked immunosorbent assay (ELISA). PAI-1 levels were found to be strongly positively correlated with the rates of relapse (P < .0001) and death (P < .001). Relating the levels of PAI-1 with those of other cytosolic prognostic factors, we found a positive association with the metastasis-related proteases uPA (P < .0001) and cathepsin D (P < .0001). On the other hand, PAI-1 levels were found to be negatively correlated with ER (P < .005) and PgR (P < .001), and the estrogen-regulated pS2-protein (P < .001), which are proteins associated with a favorable prognosis. In multivariate regression analysis for 5-year relapse-free survival, and using an optimized cutoff point for discrimination between PAI-1-positive and -negative, independent predictors of the rate of relapse were found to be PAI-1 (P < .0001) and uPA (P = .01) of the cytosolic parameters, and tumor size, lymph node status, and premenopausal age of the clinical parameters. In multivariate analysis in patients with node-negative disease, only PAI-1 (P < .001) and tumor size (P = .03) were positively and premenopausal age negatively (P < .001) associated with the rate of relapse. In patients with node-positive disease, PAI-1 (P < .001), uPA (P = .02), tumor size (P < .001), and the number of positive lymph nodes (P < .001) were all positively associated with the rate of relapse. We conclude that the PAI-1 level measured in routinely prepared cytosols is an important parameter to predict the metastatic potential in both node-negative and node-positive human primary breast cancer.