We reviewed the results of all breast core needle biopsies with a diagnosis of atypical ductal hyperplasia (ADH) or atypia not otherwise specified and subsequent excisional biopsies for a 50-month period and correlated the results. Of 3,026 biopsies, 216 were diagnosed as ADH or atypia not otherwise specified, and subsequent resection was available for 105. After review, 95 qualified as ADH. Subsequent resection showed ductal carcinoma in situ (DCIS) in 13 excisions, ADH in 31, lobular carcinoma in situ in 6, and benign proliferative lesions in the remaining 45. In none of the 8 biopsies in which DCIS was found and radiographs were available for review was the radiographic lesion entirely removed. For comparison, the incidence of carcinoma in resections done for a diagnosis of DCIS, low or intermediate grade (solid, cribriform, or micropapillary type), on core needle biopsy was significantly greater (8 of 10 cases). However, the size of the lesions diagnosed as carcinoma also was significantly greater than that of the lesions diagnosed as ADH, and in none of the 8 biopsies with DCIS at excision was the lesion entirely removed at the time of biopsy. The incidence of carcinoma in excisional biopsies done for a diagnosis of ADH in core needle biopsies in our institution is relatively low, while the incidence of ADH is relatively high. Possible reasons for this include total removal of small lesions at the time of biopsy and use of the diagnostic term ADH for lesions that are not associated with coexistent DCIS.