The objective of this one-institutional study was to determine the number of large-core needle biopsies (LCNB), under threedimensional ultrasound (3D-US) validation, that are sufficient to obtain a reliable histological diagnosis of a sonographically detectable breast lesion. Over an 28-month period, 962 sonographically guided LCNB were performed under 3D-US validation to assess 962 breast lesions. All biopsies were carried out with an automated core biopsy device fitted with 14-gauge (22 mm excursion) needles. Data of 962 biopsied breast lesions were gathered. Surgical follow-up was available for 659 lesions. Breast malignancies were diagnosed by ultrasound-guided LCNB with a sensitivity of 98.2% by performing three cores per lesion. In few cases, the open surgical specimen revealed the presence of invasive carcinomas in contrast to initial LNCB-based classification as ductal carcinomas in situ (DCIS, 11 lesions), lobular carcinoma in situ (one lesion), and atypical ductal hyperpasia (one lesion). Owing to disagreement between classification based on breast-imaging and histological findings, eight of these tumours were subsequently excised. Of the lesions that were removed at the patients' requests despite benign LCNB diagnosis, two were infiltrating carcinoma and one a DCIS. We demonstrate that three 3D-US-guided percutaneous core specimens are sufficient to achieve tissue for a reliable histological assessment of sonographically detectable breast lesions and allow the detection of malignancies with high sensitivity and low rate of false-negative diagnoses.