Prior to the current mammographic era, ductal carcinoma in situ (DCIS) usually presented as a large mass, was classified morphologically by architecture, and treated by mastectomy. The introduction of screening mammography led to an increase in the incidence of DCIS, a decrease in the average size of DCIS, and an increased emphasis on its heterogeneous nature. Thus, a reproducible and prognostically relevant classification system for DCIS is necessary. The ultimate goal of this classification is proper selection of patients for whom lumpectomy would suffice rather than mastectomy. Features to evaluate include: extent and size of disease, adequacy of resection margins, and histology. While none of the proposed histological classification systems were endorsed at the recent Consensus Conference on the Classification of DCIS, nuclear grade was the most important feature common to most of them. Architecture was given secondary importance. By definition, DCIS is a non-invasive clonal proliferation of epithelial cells originating in the terminal duct lobular unit, which would be expected to be monomorphic; however, it is the degree of nuclear pleomorphism that is primarily used to separate DCIS into low, intermediate, and high grades. Architecturally, DCIS has been divided into the following types: comedo, solid, cribriform, micropapillary, and papillary. Different architectural patterns and grades may be present in a given particular case; however, some combinations of patterns occur more frequently than others. Interobserver studies have shown nuclear grading to be interpreted with greater consistency than architecture, and nuclear grading methods have correlated with biological and molecular marker studies.