Age-specific consumption of beer, wine, and liquor was ascertained by telephone interview from 779 women in Wisconsin (United States) with newly reported diagnosis of carcinoma of the colon and rectum. Population controls (n = 2,315) interviewed for this case-control study were randomly selected from Wisconsin driver's license files and Health Care Financing Administration files. Overall, there was a modest indication that high levels of alcohol consumption (11 or more drinks per week) were associated with increased risk of large bowel cancer (adjusted odds ratio [OR] = 1.47, 95 percent confidence interval [CI] = 1.0-2.22). In site-specific analyses, only rectal cancer demonstrated a significant linear trend (P = 0.01) with increasing consumption. Significant beverage-specific effects were observed for liquor and colon cancer: the adjusted ORs for 1-2, 3-5, and 6+ drinks per week were 1.12, 1.68, 1.51, respectively (P trend = 0.01). Beer was associated significantly with rectal cancer: the adjusted ORs for 1-2, 3-5, 6-10, and 11+ drinks per week were 1.25, 1.25, 1.58, 2.42, respectively (P trend = 0.02). Wine consumption was associated inversely with these cancers. These relationships appeared to be consistent for recent, past, and total lifetime consumption, and were not attributable to differences in dietary habits.