PURPOSE: Physical activity (PA) is one of few modifiable breast cancer risk factors. There have been few studies of the relation between PA and mammographic density, especially in multiethnic populations. METHODS: In a cohort of pre- and early perimenopausal women of non-Hispanic white (N = 373), African American (N = 55), Chinese (N = 178), and Japanese (N = 166) ethnicity, we used multivariable linear regression to examine the association between two measures of mammographic density (percent density and area of density) and mutually exclusive components of recent physical activity (sports, household/caregiving and work activity, active living). RESULTS: After adjusting for race/ethnicity, menopausal status, parity, past use of hormones, body mass index, waist circumference and education, we observed nonsignificant inverse associations for percent mammographic density and the highest versus the lowest category of each of our PA domains. For example, the adjusted beta for active living = -2.62, 95% confidence interval (CI) (-5.84, 0.60). Nonsignificant inverse associations also were observed for area of density and each PA domain except work activity. However, most associations were nonlinear. CONCLUSION: Our results are consistent with a modest inverse association between multiple domains of PA and mammographic density, although findings may have been attributable to chance alone.