This study evaluated the accuracy of ultrasound-guided fine-needle aspiration cytology of the sonographically most suspicious axillary lymph node (US/FNAC) to select early breast cancer patients with three or more tumour-positive axillary lymph nodes. Between 2004 and 2014, a total of 2130 patients with histologically proven early breast cancer were evaluated and treated in the Noordwest Clinics Alkmaar. US/FNAC was performed preoperatively in all these patients. We analysed the results of US/FNAC retrospectively. Pathological axillary node status (sentinel node biopsy and/or axillary lymph node dissection) was used as reference standard. A total of 634 (29.8 %) of 2130 patients had axillary lymph node metastases on final histology. 248 node positive patients (11.6 %) had three or more positive lymph nodes. The accuracy of US/FNAC to detect three or more positive lymph nodes was 89.8 %, sensitivity was 44.8 %, specificity was 95.7 %, PPV was 58.1 %, and NPV was 92.9 %. This study shows a more than adequate accuracy of preoperative US/FNAC to detect three or more positive lymph nodes (89.8 %). However, when US/FNAC was chosen as the only axillary staging method, 6.4 % of all patients (false negative group) would have been undertreated and 3.8 % of all patients (false positive group) would have been overtreated according to the ACOSOG Z0011 criteria.