Chronic and subclinical sympathetic over-activity may underlie development of type-2 diabetes mellitus (DM). Since longer heart rate-corrected QT interval (QTc) represents predominance of sympathetic activity within autonomic balance, we investigated influences of QTc on the incidence of DM in a follow-up study. The subjects were 12,530 men and 7,163 women aged 30-59 years and apparently healthy at baseline. QTc in resting ECG was obtained using Bazett's formula (QTcB), Fridericia's formula (QTcF) and the linear regression technique (QTcLR). Incident DM was identified by 'fasting serum glucose > or =7.00 mmol l(-1) (126 mg dl(-1))' or/and 'on medication for DM'. Using Cox proportional hazard models, hazard ratio (HR) for incident DM was estimated according to the quartiles (Q1-Q4) of QTcB, QTcF or QTcLR, and its linear trends across the quartiles were checked. Baseline age, BMI, smoking, drinking, exercise and education were computed as conventional confounders. During the follow-up of 93,337 person-years for men and 51,517 person-years for women, 637 men and 192 women developed DM. The multivariate-adjusted HRs(95% confidence interval) for Q4 of QTcB, QTcF and QTcLR against corresponding Q1 were 1.85(1.46, 2.34), 1.31(1.04, 1.65) and 1.58(1.26, 1.99), respectively, for men, and 2.03(1.31, 3.16), 1.34(0.91, 2.00) and 1.58(1.04, 2.38), respectively, for women. Both sexes showed increasing trends in the HRs across the quartiles of QTcB, QTcF and QTcLR. In conclusion, QTc in resting ECG moderately but proportionally predicts the risk for development of DM in middle-aged healthy men and women. Moreover, the adverse effects of prolonged QTc are independent of those of conventional risk factors.