The results of skeletal scintigraphy performed at presentation in 465 women with histologically confirmed carcinoma of the breast were correlated with tumour size, node status, clinical course and survival during a follow-up of at least 2 years and up to 9 years. Skeletal metastases were eventually confirmed in 17.6% of the population under study, but were identified at presentation by scintigraphy in only half of these. The incidence of significant scintigraphic abnormalities ranged from 1.5% in patients with T0 and T1 node negative tumours to 20.3% in T4 node positive tumours. Patients with scintigraphic evidence of metastases had a significantly shorter survival than those without; 13.6% of the patients with an abnormality considered to be significant on the criteria employed in this study failed to develop confirmatory evidence of skeletal metastases during the period of follow-up. Alternative criteria are proposed which would substantially decrease the incidence of false positives without increasing the incidence of false negatives. Even so, it is concluded that there is no evidence that routine skeletal scintigraphy affects management of newly diagnosed patients with carcinoma of the breast. Unless an algorithm can be defined which requires the information provided by skeletal scintigraphy in order to determine clinical management, this investigation is not justified as a routine staging procedure and should be reserved for patients in whom there is a clinical suspicion of metastases and for clinical trials.