This retrospective study aimed (1) to compare the diagnostic accuracy of whole-body FDG PET/CT for initial breast cancer staging with the accuracy of a conventional, multimodal imaging algorithm, and (2) to assess potential alteration in patient management based on the FDG PET/CT findings. Patients with primary breast cancer (106 women, mean age 57 ± 13 years) underwent whole-body FDG PET/CT and conventional imaging (X-ray mammography, MR mammography, chest plain radiography, bone scintigraphy and breast, axillary and liver ultrasonography). The diagnostic accuracies of FDG PET/CT and a conventional algorithm were compared. Diagnostic accuracy was assessed in terms of primary tumour detection rate, correct assessment of primary lesion focality, T stage and the detection rates for lymph node and distant metastases. Histopathology, imaging or clinical follow-up served as the standards of reference. FDG PET/CT was significantly more accurate for detecting axillary lymph node and distant metastases (p = 0.0125 and p < 0.005, respectively). No significant differences were detected for other parameters. Synchronous tumours or locoregional extraaxillary lymph node or distant metastases were detected in 14 patients (13%) solely by FDG PET/CT. Management of 15 patients (14%) was altered based on the FDG PET/CT findings, including 3 patients with axillary lymph node metastases, 5 patients with extraaxillary lymph node metastases, 4 patients with distant metastases and 3 patients with synchronous malignancies. Full-dose, intravenous contrast-enhanced FDG PET/CT was more accurate than conventional imaging for initial breast cancer staging due to the higher detection rate of metastases and synchronous tumours, although the study had several limitations including a retrospective design, a possible selection bias and a relevant false-positive rate for the detection of axillary lymph node metastases. FDG PET/CT resulted in a change of treatment in a substantial proportion of patients.