Objectives : To determine the relationship between exercise and the incidence of type 2 diabetes in an impaired fasting glucose group. Methods : This prospective cohort study was conducted in 19,440 men and 4,297 women, aged 30-69 years, with impaired fasting glucose at baseline who had undergone biennial medical evaluation through the National Health Insurance Corporation from 2000 to 2004. Impaired fasting glucose was defined as fasting glucose of 100 to 125 mg/dl and the subjects were divided into 3 groups depending on weekly exercise frequency. Multivariate logistic regression analysis was used to evaluate the association between the baseline exercise status and incidence of type 2 diabetes. Results : During the 4-year follow-up, a total of 3,239 men and 283 women developed type 2 diabetes, a cumulative incidence of 16.6% for men, and 6.5% for women. Also, 1,688 men (21.2%) and 127 women (15.2%) developed type 2 diabetes in the obese group. The adjusted relative risk (RR) of developing type 2 diabetes in non-exercising men was significantly higher than exercising men regularly (RR= 1.375, 95% CI=1.236-1.529)(p<0.0001), and the RR for non-exercising women was higher than exercising women regularly (RR=1.124, 95% CI=0.711-1.778). The RR for non-exercise men/women in the obese group was 1.571 (95% CI=1.351-1.827)(p<0.0001)/1.869 (95% CI=0.846-4.130). Conclusions : Regular exercise is effective in preventing type 2 diabetes in people with impaired fasting glucose, and particularly in obese people. People with risk factors for diabetes should participate in a regular exercise program.