The frequency with which ductal carcinoma in situ (DCIS) is detected has increased greatly since the introduction of mammographic screening. The number of treatment options has also increased and mastectomy has been extensively replaced by local excision with or without radiotherapy. DCIS is generally unicentric, as evidenced by the rarity with which it is bilateral and the location of recurrences at the site of previous surgery. Complete excision is thus curative but assessing adequacy of excision is beset with significant technical problems and consequently margin involvement does not correlate very well with the presence of residual disease in the breast or the development of clinical recurrence. Lesion size is related to recurrence but is also often difficult to measure. At the histological level, DCIS is a heterogeneous group of proliferations varying in cytological and architectural features, some of which are related to clinical outcome. The traditional method of classification was by growth pattern but was found to lack reproducibility and prognostic power. As a consequence, several new classifications have been proposed in recent years. Some have been assessed more rigorously than others in terms of the consistency with which they can be applied and their ability to predict clinical outcome. There is strong evidence, however, that nuclear grade is the best predictor of recurrence and the time scale over which it is likely to occur although presently it can be determined with only fair to moderate consistency. Necrosis is also a useful feature when used in combination with nuclear grade, but specifically recognizing a comedo pattern appears to have little clinical value and is associated with significant diagnostic inconsistency. No histological features to date have been found to predict the development of invasive disease. Histological assessment alone is insufficient to determine how patients with DCIS should be managed, which should also take account pathological assessment of excision margins and lesion size as well as radiological and clinical features.