BACKGROUND: Diabetes is marked by metabolic dysregulation and high cardiovascular risk. Preceding diabetes onset, cardiac change markers often appear, yet predicting occurrence of diabetes remains challenging. METHODS: The study included 11,297 ARIC Study participants aged 45-64 without baseline diabetes or heart disease, with 12‑lead ECGs recorded. Cox regression models were used to analyze repolarization parameters (QTc intervals, J-point level, T-wave amplitude, T-wave to R-wave (T/R) ratio) in relation to diabetes risk. Correlation analyses explored links between ECG findings, metabolic parameters, and myocardial fat content. RESULTS: Over 23 years, 3338 participants (29.5 %) developed type 2 diabetes. A reduced T/R ratio showed dose-response relationship with diabetes risk, notably in lead I (HR 1.22, 95 % CI 1.15-1.31) and lead V5 (HR 1.15, 95 % CI 1.07-1.24) per standard deviation (SD) decrease after adjustment for common covariates. J-point level and T-wave amplitude also exhibited associations, though weaker. Negative correlations were found between repolarization markers, including T-wave amplitude, T/R ratio with metabolic parameters. CONCLUSIONS: ECG parameters, especially T-wave amplitude and T/R ratio, predict incident type 2 diabetes and serve as potential early biomarkers for metabolic-induced cardiac change. These findings underscore their clinical relevance in identifying individuals at risk for diabetes.