Purpose: Previous studies have shown that bone metastases are revealed by magnetic resonance imaging (MRI) or bone marrow scintigraphy several months before they are visible by conventional bone scintigraphy (BS). We present a new approach for detecting bone metastases in patients with breast cancer. We compared findings obtained with fluoride ion (F-18) and positron emission tomography (PET) with those obtained with conventional BS. Patients and Methods: Thirty-four breast cancer patients were prospectively examined using F-18-PET and conventional BS. F-18-PET and BS were performed within 3 weeks of each other. Metastatic bone disease was previously known to be present in six patients and was suspected (bone pain or increasing levels of tumor markers, Ca²⁺, alkaline phosphatase) in 28 patients. Both imaging modalities were compared by patient-by-patient analysis and lesion-by-lesion analysis, using a five-point scale for receiver operating characteristic (ROC) curve analysis. A panel of reference methods was used, including MRI (28 patients), planar x-ray (17 patients), and spiral computed tomography (four patients). Results: With F-18-PET, 64 bone metastases were detected in 17 patients. Only 29 metastases were detected in 11 patients with BS. As a result of F-18-PET imaging, clinical management was changed in four patients (11.7%). For F-18-PET, the area under the ROC curve was 0.99 on a lesion basis (for BS, it was 0.74; P < .05) and 1.00 on a patient basis (for BS, it was 0.82; P < .05). Conclusion: F-18-PET demonstrates a very early bone reaction when small bone marrow metastases are present, allowing accurate detection of breast cancer bone metastases. This accurate detection has a significant effect on clinical management, compared with the effect on management brought about by detection with conventional BS.