The purpose of this paper is to present background information on carcinoma in situ (CIS) of the breast and to provide a theoretical framework for planning epidemiologic studies which may further our understanding of breast cancer. Two types of epidemiologic studies are needed which incorporate CIS of the breast: (i) case-control studies, in which in-situ lesions serve as disease outcomes (endpoints), and (ii) cohort studies and clinical trials, in which diagnosis of in-situ carcinoma serves as a starting point for patient treatment and follow-up. Case-control studies focusing on the causes of CIS have distinct advantages: if risk factors for cancer contribute to pathways involving some intermediate stages but not others (e.g. comedo-type but not non-comedo-type DCIS; LCIS versus DCIS), the use of precursor lesions may more clearly reveal risk factor associations than studies of invasive breast cancer alone; epidemiologic studies of precursor lesions are conducted closer in time to the exposures suspected to be causes and may reduce recall bias or other forms of misclassification; genetic alterations in early lesions are more likely to represent causal events in development of the malignant phenotype. Population-based case-control studies of CIS may thus prove useful in understanding breast cancer etiology and designing preventive strategies. CIS patients identified for case-control studies may be followed up over time as a cohort. Cohort studies (and clinical trials) of CIS aim to elucidate mechanisms influencing progression of CIS to invasive cancer as well as to evaluate effectiveness of specific treatment modalities. Although the majority of CIS lesions of the breast are ductal carcinoma in situ (DCIS), epidemiologic studies which also include patients with lobular carcinoma in situ (LCIS) address potential differences between DCIS and LCIS with respect to both etiology and progression.