In a prospective study of 526 consecutive patients with operable breast cancer, the significance of positive parathyroid hormone-related protein (PTHrP) staining by immunohistology has been evaluated for a median of 10-year follow-up. Improved survival was observed for the 79% of tumors which stained positively for PTHrP [estimated univariate hazard ratio, 0.43; 95% confidence interval (95% CI), 0.30-0.62; P < 0.001]. Adjustments for N stage, progesterone receptor status, and log tumor size changed this estimate only slightly to 0.47 (95% CI, 0.63-0.69; P = 0.001). Patients with PTHrP-positive primary tumors were less likely to develop bone metastases (hazard ratio, 0.63; 95% CI, 0.41-0.98; P = 0.04). PTHrP status was associated with estrogen receptor (P = 0.01), progesterone receptor (P = 0.03), and menopausal status (P = 0.006) but was not significantly associated with tumor size, vascular invasion, tumor grade, or patient age. Of 19 patients requiring surgery for bone metastases, the primary cancers were PTHrP negative in seven, all but one of whom had PTHrP-positive bone metastases. All 12 patients with PTHrP-positive primary cancers also had positive bone metastases. We conclude that increased production of PTHrP by breast cancers confers on them a less invasive phenotype, an effect distinct from the bone resorption-stimulating action that favors bone metastasis. It is likely that the latter property is influenced by factors in the bone microenvironment.