To explore whether the triglyceride to high-density lipoprotein cholesterol ratio (TG/HDL-C) was independently associated with the risk of incident type 2 diabetes mellitus (T2DM) in a large Japanese cohort. A secondary analysis was performed using open-access data from a retrospective cohort study. A total of 12,716 eligible participants who had standard medical examinations at the Murakami Memorial Hospital were included in this study. New-onset T2DM was the main outcome during follow-up. The risk of T2DM based on the TG/HDL-C ratio was evaluated using Cox regression analysis and Kaplan-Meier analysis. Subgroup analysis was performed to understand further the significance of the TG/HDL-C ratio in particular populations. To assess the potential of the TG/HDL-C ratio for predicting T2DM, a receiver operating characteristic (ROC) analysis was performed. One hundred fifty new-onset T2DM cases were observed during a median follow-up of 5.39 years. The incidence of T2DM increased with a rise in the TG/HDL-C ratio based on the Kaplan-Meier curves ( < 0.0001). After controlling for potential confounding variables, the TG/HDL-C ratio was positively related to incidence of T2DM (hazard ratio = 1.08, 95% confidence interval: 1.01-1.15,  = 0.0239). In subgroup analysis, those with a body mass index of ≥18.5 and <24 kg/m showed a significantly positive relationship. The area under the ROC curve for the TG/HDL-C ratio as a T2DM predictor was 0.684. The optimal TG/HDL-C ratio cutoff value for T2DM was 1.609, with a sensitivity of 54.7% and a specificity of 73.6%. The authors' results showed a significant relationship between the TG/HDL-C ratio and the incidence of T2DM in the Japanese population.