A multiparametric biological study of 178 evaluable breast cancer patients was performed to evaluate the usefulness of individual parameters for the prediction, diagnosis and follow-up of bone metastasis (BM) as compared to physical tests. Serum and urinary calcium and phosphorus were of little discriminative value. The sensitivity and specificity for the urinary hydroxyproline/creatinine ratio (UHP/creat.) were respectively 75% and 81.4% versus 43.2% and 91.4% for the bone isoenzyme of alkaline phosphatase in serum (BIE). During an average follow-up of 14.3 months for 51 patients with bone metastasis, variations in UHP-creat. and BIE correlated with clinical and physical findings; the best correlation was given by UHP/creat. Twenty patients who were initially free of bone metastasis were followed-up for a mean metastasis-free interval of 14 months; UHP/creat. predicted the appearance of bone metastasis in 50% of cases with a lead time of 5.5 months versus 40% and 8.5 months for BIE. Combined use of UHP/creat. and BIE gave a prediction rate of 70% and a mean lead time of 7 months, and this combination of markers thus merits consideration in the management of breast cancer patients.