We investigated the possibility that cathepsin D, an estrogen-induced lysosomal protease, might have value as a prognostic factor in breast cancer by studying frozen tissue specimens from 397 patients. We measured the 34-kd mature form of the enzyme by Western blot assay and densitometry. Among 199 patients with node-negative disease, but not among 198 with node-positive disease, high levels of cathepsin D proved to be a significant predictor of reduced disease-free survival (median follow-up, 64 months), either as a continuous variable (log cathepsin D; P = 0.018) or as a dichotomous variable with an optimized cutoff point (P = 0.0001). Results were similar for overall survival (P = 0.009 and 0.0001, respectively). Relating the level of cathepsin D to other prognostic factors in the patients with node-negative disease, we found an association with aneuploidy but none with estrogen or progesterone receptors, tumor size, or the age of the patient. In multivariate analyses, a high level of cathepsin D was the most important independent factor in predicting shorter disease-free and overall survival in patients with node-negative disease. As compared with the risk in women with low levels of cathepsin D, the relative risk of tumor recurrence was 2.6 (95 percent confidence interval, 1.6 to 4.4) and the relative risk of death was 3.9 (95 percent confidence interval, 2.1 to 7.3) among those with high levels of cathepsin D. For disease-free survival, cathepsin D status was predictive of outcome primarily among those with aneuploid tumors; the actuarial five-year recurrence rates of aneuploid tumors were 60 percent among women with high levels of cathepsin D and 29 percent among those with low levels, as compared with 22 percent for all diploid tumors. We conclude that cathepsin D may be an independent predictor of early recurrence and death in node-negative breast cancer.