Data on known risk factors for human breast cancer are reviewed. The major predictors of risk are age, geographic area of residence, age at first birth, certain indicators of ovarian activity, history of benign breast disease, and familial history of breast cancer. Specific etiologic hypotheses are also reviewed, with consideration of the extent to which they explain, or are consistent with, the observed patterns of risk distribution. If a human mammary tumor virus exists, exposure is prpbably widespread, and other causal factors must account for the observed risk patterns. Excretion of low levels of androgen metabolites is associated with increased risk of breast cancer, but probably is only an indicator of other unidentified endocrine patterns. A hypothesis linking breast cancer risk to a particular pattern of estrogen metabolism in early reproductive life, although unsupported by direct evidence, is compatible with most epidemiologic and experimental observations on the disease. The roles of exogenous estrogens, prolactin, and progesterone—suggested by animal models-remain essentially unevaluated in humans. A small number of human breast cancers are probably caused by exposureto high levels of ionizing radiation. Familial factors seem most strongly implicated in young women with bilateral disease, but even among such patients the nature of the familial factors, genetic or environmental, is unknown.