OBJECTIVE - To determine if impaired fasting glucose (IFG; fasting plasma glucose level 6.1-6.9 mmol/l) can predict future type 2 diabetes as accurately as does impaired glucose tolerance (IGT; 2-h plasma glucose level 7.8-11.0 mmol/l). RESEARCH DESIGN AND METHODS - A longitudinal population- based study was performed with surveys in 1987 and 1992 on the island of Mauritius, assessing diabetes status by the oral glucose tolerance test. A total of 3,717 subjects took part in both surveys. Of these subjects, 3,229 were not diabetic in 1987 and formed the basis of this study. RESULTS - At baseline, there were 607 subjects with IGT and 266 subjects with IFG. There were 297 subjects who developed diabetes by 1992. For predicting progression to type 2 diabetes, the sensitivity, specificity, and positive predictive values were 26, 94, and 29% for IFG and 50, 84, and 24% for IGT, respectively. Only 26% of subjects that progressed to type 2 diabetes were predicted by their IFG values, but a further 35% could be identified by also considering IGT. The sensitivities were 24% for IFG and 37% for IGT in men and 26% for IFG and 66% for IGT in women, respectively. CONCLUSIONS - These data demonstrate the higher sensitivity of IGT over IFG for predicting progression to type 2 diabetes. Screening by the criteria for IFG alone would identify fewer people who subsequently progress to type 2 diabetes than would the oral glucose tolerance test.