Background: Dietary factors are well-known modifiable risk factors for type 2 diabetes (T2D), but many studies overlook the interrelationships between these factors, even though foods are often consumed together and contain a variety of nutrients. Objectives: In this study, we employed a diet-wide association study approach to investigate the links between various dietary factors and T2D onset, taking into account complex dietary patterns. Methods: We analyzed 16,666 participants without T2D from three Korean population-based cohorts: the Multi-Rural Communities Cohort (n = 8302), the Atherosclerosis Risk of a Rural Area Korean General Population cohort (n = 4990), and the Kanghwa cohort (n = 3374). A two-step approach was employed. In the first step, robust Poisson regression analysis was used for the initial screening (false discovery rate-adjusted p-values < 0.05). In the second step, a hierarchical cluster analysis was conducted of all dietary factors, followed by mutual adjustment of the screened factors within each cluster to account for interrelationships. Results: The 11 food clusters screened were cooked rice with beans, rice cakes, breads/spreads, bread products, cheese and pizza/hamburger, grain powder, snack/confections, nuts and roasted beans, soy milk, traditional beverages, and non-native fruit. These factors were similarly distributed across three of the seven clusters in each cohort. After mutual adjustment, cooked rice with beans (p-value ≤ 2.00 × 10−7 in all three cohorts) and non-native fruits (p-value ≤ 5.91 × 10−3 in two cohorts) remained significantly associated with lower T2D risk in more than one cohort. Conclusions: The inverse association of cooked rice with beans, not observed with other types of cooked rice, and that of non-native fruits, suggest that incorporating beans into rice and eating various fruits may be an effective strategy for preventing diabetes.