In October 2002, ultrasound-guided, vacuum-assisted, percutaneous excision was shown to facilitate the complete removal of benign breast lesions up to 3 cm in diameter. This study was performed to ascertain the overall accuracy of ultrasound-guided, vacuum-assisted, percutaneous excision as evidenced by the frequency of atypical ductal hyperplasia (ADH) underestimation. A retrospective review was conducted of 542 consecutive ultrasound-guided, vacuum-assisted breast biopsies performed between February 2000 and September 2004. Before July 2002, no attempt was made to completely remove all imaged lesion evidence. After July 2002, all patients underwent complete percutaneous excision of all imaged lesion evidence. Pathology review revealed 52 lesions that demonstrated ADH and no evidence of malignancy. Each patient with this diagnosis was offered surgical excision. Pathologic reports for each group were compared with the subsequent open surgical specimens. Of 542 consecutively diagnosed lesions, 52 displayed ADH with no evidence of malignancy (10%). Five patients refused operation. Of the 47 patients who underwent open excision, 6 (13%) were found to have malignancies. The rate of ADH underestimation was 6 of 18 (33%) in incisional biopsies and 0 of 29 performed with complete imaged lesion evidence (p=0.002). The rate of ADH underestimation in women who underwent ultrasound-guided, vacuum-assisted, percutaneous excision was zero, a result equivalent to open surgical biopsy. ADH is a more common finding in sonographic lesions than has been previously reported. Complete ultrasound-guided, vacuum-assisted, percutaneous excision is more accurate than nonexcisional ultrasound-guided biopsy. Patients so diagnosed have very low underestimation rates and may not require open surgical reexcision.