The evaluation of mammographic abnormalities has become a substantial effort for surgeons and radiologists. The vacuum-assisted core biopsy (VACB) has been touted as a more accurate tool for the evaluation of mammographic lesions. Diagnosis of atypical ductal hyperplasia (ADH) from a percutaneous needle biopsy of the breast is associated with a significant risk of missing a significant breast lesion. We compared 2 methods of sampling with stereotactic-guided breast biopsy, 14-gauge automated gun core biopsy (AGCB) and VACB, on the accuracy of diagnosis of ADH at a single institution. All cases of ADH, without associated malignancy, found via image-guided breast biopsy of nonpalpable lesions between March 1996 and April 2002 were evaluated. VACB biopsy needles were utilized between July 1998 to April 2002 (686 patients) and 14-gauge AGCB from March 1996 to June 1998 (350 patients). The results of these biopsies were reviewed and compared to surgical biopsy and pathological records. ADH alone was found in 53 cases (5.1% of biopsies; mean age 57.9 years). Of these, 39 patients with ADH subsequently underwent wire-localized excisional biopsy. The other 14 patients were observed. VACB biopsy understaged 7 of 29 (24%) patients with ADH (all of which were DCIS), AGCB understaged 4 of 10 cases (40%) with one being invasive. Of the patients in the core biopsy group who were initially followed, 2 developed significant lesions within 3 years of follow-up in the same quadrant of the breast. If these cases are added to the AGCB group, then 50 per cent were understaged and significantly more invasive lesions were understaged than with VACB (17% vs. 0%; P = 0.018). The VACB resulted in less understaging of ADH than AGCB. However, there remains a significant risk of missing DCIS in this setting even with the VACB. Furthermore, the risk of understaging an invasive lesion is significantly lower in this setting with a VACB than an AGCB. Although the risk of understaging ADH is lower with the VACB, we continue to recommend excisional biopsy in a good-risk patient when a diagnosis of ADH is rendered via VACB biopsy.