In this article, the probability of finding malignancy on surgical excision after applying well-defined morphological criteria combined with immunohistochemical evaluation of cytokeratin 5/6 for the diagnosis of atypical ductal hyperplasia on core biopsies is examined. On the basis of morphology alone, the reviewers reclassified the diagnoses of 140 core biopsies as follows: atypical ductal hyperplasia (n = 64), ductal hyperplasia of usual type (n = 44), flat epithelial atypia (n = 11), and miscellaneous benign (n = 21). Cytokeratin 5/6 immunostain was negative in 85.7% of atypical ductal hyperplasia cases and positive in 77.8% of ductal hyperplasia of usual type cases. The probability of predicting malignancy in a surgical specimen following a core biopsy increased from 43.6% to 67.8% (P = .002) by adhering to defined criteria and using cytokeratin 5/6 immunostain. Expertise and adherence to defined criteria are required to establish an accurate diagnosis of atypical ductal hyperplasia. Cytokeratin 5/6 can be a useful adjunct in cases with ductal hyperplasia but not in columnar cell lesions, where it is universally negative.