The correlation between response of metastatic bone lysis and bone pain, various biochemical markers of bone metabolism, and radiological and scintigraphic findings was investigated in 49 women with breast cancer included in a calcitonin supportive therapy trial. All patients had dominant skeletal disease and were on stable systemic treatment (cytotoxic or hormonal) for a least 6 months before the first response evaluation. Bone pain correlated poorly with treatment response. Changes in sclerotic metastases did not show any apparent relation to changes in lytic lesions. A correlation between bone scans and lytic activity on radiographs was found. The absolute level of biochemical bone markers did not correlate with treatment response, but seemed instead to reflect the rate of bone turnover. The relative level of bone markers with respect to baseline showed better correlation to treatment response. The best conventional marker of disease activity was urinary hydroxyproline/creatinine. Propeptide of Type III procollagen (PIIINP), a novel marker reflecting collagen turnover, promises to be at least as sensitive as hydroxyproline. Stable and regressing patients had the same prognosis and significantly longer survival than progressors.