Major dietary patterns were studied for the ability to predict type 2 diabetes mellitus in a cohort of 4,304 Finnish men and women aged 40-69 years and free of diabetes at baseline in 1967-1972. Factor analysis was used to identify dietary patterns from dietary data that were collected using a 1-year dietary history interview. A total of 383 incident cases of type 2 diabetes occurred during a 23-year follow-up. Two major dietary patterns were identified. The pattern labeled "prudent" was characterized by higher consumption of fruits and vegetables, and the pattern labeled "conservative" was characterized by consumption of butter, potatoes, and whole milk. The relative risks (adjusted for nondietary confounders) between the extreme quartiles of the pattern scores were 0.72 (95% confidence interval: 0.53, 0.97; p(trend) = 0.03) for the prudent pattern and 1.49 (95% confidence interval: 1.11, 2.00; p(trend) = 0.01) for the conservative pattern. Thus, the prudent dietary pattern score was associated with a reduced risk and the conservative pattern score was associated with an increased risk of type 2 diabetes. In light of these results, it appears conceivable that the risk of developing type 2 diabetes can be reduced by changing dietary patterns.