Twenty-seven patients taking part in an advanced breast cancer trial in Australia and New Zealand were studied to assess the predictive value of serial bone scans as an indicator of response to therapy. The report of serial bone scans in 20 of the 27 patients correctly reflected the clinical outcome. Of the 20 patients, all had an initial subjective response and 15 had an objective response. Of the seven patients whose bone scan report did not reflect the clinical situation, five had increased uptake in known lesions and new lesions and one had increased uptake in known lesions only within six months of commencing therapy. These scan reports indicated progression of disease rather than a healing or "flare" response which was later proved to exist. One patient had clinical progression of bone metastases which was confirmed by x-ray but was not reflected in the bone scan until months later. Clinical progression of disease was accurately reflected in the bone scans of 11 of 12 patients. Recognition that a "'flare" effect in the first months of therapy may also cause new lesions to be seen should enhance the accuracy of bone scintigraphy.