BACKGROUND: Ductal carcinoma in situ represents about 20% of all tumours diagnosed within mammographic screening programs. The natural history of DCIS is poorly understood, as it cannot be observed directly. Estimates of the proportion of DCIS that progress to invasive cancer, as well as factors that may influence progression, are important for clinical management. Here we review various sources of evidence regarding the natural history of DCIS. METHODS: We identified relevant publications of studies on: follow-up studies of DCIS initially misdiagnosed as benign, studies of recurrence of DCIS as invasive cancer, autopsy studies, studies of risk factors for DCIS, animal studies and studies that used mathematical models to study growth of DCIS and invasive cancer. Data sources included the MEDLINE data base, searches of articles cited in key reviews and editorials. RESULTS: The most direct evidence regarding the progression of DCIS to invasive cancer comes from studies where DCIS was initially misdiagnosed as benign and treated by biopsy alone. These studies suggest that between 14-53% of DCIS may progress to invasive cancer over a period of 10 or more years. The reported prevalence of undiagnosed DCIS in autopsy studies, of approximately 9%, has been used to suggest a larger reservoir of DCIS may exist in the population. All types of study designs reviewed had limitations that may bias the estimate of progression in either direction. CONCLUSION: The available evidence suggests not all DCIS will progress to invasive cancer in the medium term but precise estimates of progression are not possible given the limitations of the data. Mathematical modelling of various scenarios of progression and studies of genetic factors involved in progression may shed further light on the natural history of DCIS.