Of the limited number of epidemiological investigations on aspirin (and other nonsteroidal anti-inflammatory drugs) and breast cancer, the majority observe a protective role, yet only a few report dose-response effects for frequency or duration of use. We studied aspirin use among 1,478 breast cancer patients diagnosed from 1982 to 1998, and 3,383 cancer-free hospital controls at the Roswell Park Cancer Institute. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were estimated using unconditional logistic regression. Compared to never use,both regular (> or =1 tablet per week for > or =1 year) and occasional use were inversely associated with breast cancer (adjusted OR = 0.84, 95% CI 0.64-0.97; adjusted OR = 0.80, 95% CI 0.67-0.96, respectively). Among regular users, an inverse trend was found for number of tablets consumed per week (1, 2-6, or > or =7) with corresponding ORs of 0.95, 0.80, and 0.74 (P(trend) = 0.01). Daily use spanning 10 or more years was associated with a more pronounced reduction in risk (P(trend) = 0.005). Our findings corroborate the growing body of observational evidence that regular aspirin use may be associated with reduced risk of breast cancer.