BackgroundTo assess the accuracy of ultrasound-guided 16G or 18G core needle biopsy (CNB) for ultrasound-visible breast lesions, and to analyze the effects of lesion features.MethodsBetween July 2005 and July 2012, 4,453 ultrasound-detected breast lesions underwent ultrasound-guided CNB and were retrospectively reviewed. Surgical excision was performed for 955 lesions (566 with 16G CNB and 389 with 18G CNB) which constitute the basis of the study. Histological findings were compared between the ultrasound-guided CNB and the surgical excision to determine sensitivity, false-negative rate, agreement rate, and underestimation rate, according to different lesion features.ResultsFinal pathological results were malignant in 84.1% (invasive carcinoma, ductal carcinoma in situ, lymphoma, and metastases), high-risk in 8.4% (atypical lesions, papillary lesions, and phyllodes tumors), and benign in 7.5%. False-negative rates were 1.4% for 16G and 18G CNB. Agreement rates between histological findings of CNB and surgery were 92.4% for 16G and 92.8% for 18G CNB. Overall underestimate rates (high-risk CNB becoming malignant on surgery and ductal carcinoma in situ becoming invasive carcinoma) were 47.4% for 16G and 48.9% for 18G CNB. Agreements were better for mass lesions (16G: 92.7%; 18G: 93.7%) than for non-mass lesions (16G, 85.7%; 18G, 78.3%) (P <0.01). For mass lesions with a diameter ≤10 mm, the agreement rates (16G, 83.3%; 18G, 86.7%) were lower (P <0.01).ConclusionsUltrasound-guided 16G and 18G CNB are accurate for evaluating ultrasound-visible breast mass lesions with a diameter >10 mm.