Ductal carcinoma in situ (DCIS) constitutes a major public health problem, with up to half of screen-detected cancers representing pure forms of DCIS without evidence of invasion. A proportion of cases detected with routine screening would not have progressed to a life-threatening form of breast cancer during the patient’s lifetime, and overdiagnosis of breast cancer is a cause for concern. Once DCIS has been detected, treatment is obligatory and present technologies do not allow accurate risk stratification such that intensity of treatment can be tailored to risk of recurrence and progression to invasive disease. Present management strategies are based on prognostic and predictive information derived from conventional histopathological and host factors. With increasing molecular characterisation of these preinvasive lesions, data will be available for how factors such as oestrogen receptor, progesterone receptor, HER2, and indicators of proliferative activity can provide additional information about both prognosis and benefit from adjuvant treatments such as radiotherapy and hormonal therapy. Low-risk patients are especially poorly defined in terms of need for adjuvant therapies, which can be associated with both short-term adverse sequelae and long-term effects (eg, cardiotoxicity) that can affect all-cause mortality. Optimum risk prediction in the future is likely to be achieved by integration of both conventional and molecular factors, which should be incorporated into a validated predictive model to help with clinical decision making.