A workshop on physical activity and breast cancer was held in November 1997 to review previous epidemiologic research on this topic and to identify new areas for research. This article is the first of three summaries of the workshop's activities. The material reviewed included 21 studies that reported a measure of physical activity in relation to breast cancer outcomes and were published by December 1997. They were identified in a computerized literature search and a "by-hand" review of journals. The study designs, populations, data collection methods, and results were examined and the strengths and limitations of the studies identified. The strengths and limitations are discussed herein, as are recommendations for future research. Fifteen of the 21 studies suggested that physical activity reduces the risk of breast cancer, whereas four studies found no association and two studies found an increased risk of breast cancer associated with physical activity. Specific subgroups of the population may experience a greater decrease in breast cancer with increased levels of physical activity. These include women who are lean, parous, and premenopausal. Some examination of confounding and effect modification was undertaken. Hypothesized biologic mechanisms for this putative association include an effect of physical activity on endogenous hormones, energy balance, and the immune system. The overall evidence supports a reduction in breast cancer risk with increased physical activity. However, numerous questions remain regarding this putative association. These include the underlying biologic model and the parameters of physical activity that are associated with risk, such as the types of activity (occupational, recreational, and household), the components of activity (frequency, intensity, and duration), the time periods in life that are associated with risk reduction, and the important confounders and effect modifiers of this association. Use of intermediate endpoints for breast cancer may be useful in such investigations.