BACKGROUND: The sensitivity of axillary ultrasonography (AUS) has increased in recent years, enabling detection of even low-volume axillary nodal metastases. The aim here was to evaluate the axillary tumour burden in women with a fine-needle aspiration/core biopsy-proven positive node on AUS and in those with a positive sentinel node biopsy (SNB). METHODS: This retrospective cohort study included all patients with early breast cancer who had AUS and axillary lymph node dissection (ALND) between 2011 and 2014. RESULTS: A total of 332 patients who had ALND were eligible for the study, 191 (57·5 per cent) in the AUS-positive group and 141 (42·5 per cent) in the SNB-positive group. Patients in the AUS-positive group were older at diagnosis (P = 0·018), more likely to have larger tumours (P = 0·002), higher tumour grade (P = 0·005), positive human epidermal growth factor 2 status (P = 0·015), and negative oestrogen receptor status (P < 0·001). The AUS-positive group also had a larger number of lymph nodes with macrometastases (P < 0·001) and were more likely to have extranodal invasion (P < 0·001). In the AUS-positive group, 40·3 per cent of patients (77 of 191) had only one or two nodes with macrometastases identified at histology after ALND. Tumour size no larger than 20 mm, invasive ductal or lobular histology and breast-conserving surgery were associated with the presence of two or fewer macrometastases at ALND. Only tumour size and tumour histology remained significant in multiple logistic regression analysis. CONCLUSION: Patients with AUS-detected metastases had a higher axillary tumour burden than those with SNB-detected metastases. Around 40 per cent of patients with AUS-detected nodal disease had one or two nodes with macrometastases and were thus overtreated by ALND.