Menstrual characteristics may serve as surrogate measures of endogenous estrogen and may be related to breast cancer risk. No previous studies have systematically investigated menstrual factors in relation to the disease in African-American women. This case-control study is aimed to assess the relationship between menstrual factors and breast cancer in African-American women. Cases were 304 African-American women, aged 20-64 living in three Tennessee counties, diagnosed with breast cancer between 1995 and 1998. Controls were selected through random-digit dialing and frequency matched to cases (n=305). Phone interviews were conducted on menstrual factors--age at menarche, time to regularity, cycle length, flow length, age at menopause--and other risk factors. Logistic regression showed that compared to women with short cycle length (<28 days), women with average cycle length > or =28 had decreased risk of breast cancer (odds ratio (OR)=0.60, 95% confidence interval (CI), 0.38-0.94). Dose-response analyses showed decreasing risk with longer cycle length. Results by menopausal status revealed an inverse relationship was shown only in postmenopausal women. No significant associations were observed for other menstrual factors. Findings suggest that cycle length has an inverse association with breast cancer in African-American women that may primarily exist for post-menopausal tumors.