A case-control study was conducted in Ibaraki, Japan. Cases were 148 women aged 26-69 diagnosed with breast cancer at Tsukuba University Hospital or Tsukuba Medical Center Hospital between January 1990 and March 1997. Two controls were individually matched to cases by age (within 1 year) and residence (municipalities). A self-administered questionnaire was used to obtain information on physical activity and the effects of potential confounding factors. Conditional logistic regression analysis was used to estimate odds ratios (ORs) and 95 percent confidence intervals (95% CIs). After adjustment for potential confounding factors (height, body mass index (BMI), family history of breast cancer, education, age at menarche, age at first birth, parity, menopausal status), recreational physical activity was associated with a reduced risk of breast cancer (p for trend = 0.005). OR for breast cancer among women who played regular sport or exercise more than 15.3 METs per week was 0.35 (95% CI = 0.17, 0.73), as compared to women with no sport or exercise. The risk of breast cancer was also reduced in women with higher occupational physical activity levels, but it was not significant (p for trend = 0.223). The adjusted OR among the highest quartile level was 0.55 (95% CI = 0.27, 1.12) with the lowest quartile level as reference. Our findings support a protective effect of recreational physical activity on breast cancer in Japanese women.