In the time period 1996-2004, a case-control study of colorectal cancer was conducted in Montevideo, Uruguay. The study included 610 cases and 1,220 controls, frequency matched for age, sex, and residence. All cases were newly diagnosed and microscopically confirmed and controls were drawn from the same hospitals. Controls were submitted to factor analysis (principal components method) and 4 dietary patterns for men (prudent, traditional, Western, drinker) and 3 for women (prudent, Western, drinker) were retained. These were rotated and normalized by the Kaiser method. Scores were applied to all participants (cases and controls) and odds ratios were estimated by logistic regression and polynomial regression. The Western pattern showed an OR of 2.62 (95 % CI 1.36-5.08) for colon cancer among men, and women displayed a similar increase in risk. However, rectal cancer was not associated with this diet, rather being inversely associated with the prudent and traditional patterns among men (OR 0.49, 95 % CI 0.28-0.57 for the traditional pattern). In conclusion, whereas the Western pattern was directly associated with colon cancer, the prudent pattern was strongly protective for rectal cancer.