Non-alcoholic fatty liver disease (NAFLD) is associated with a high risk of type 2 diabetes (DM), therefore, early diagnosis of NAFLD is important to prevent incident DM. FIB-4 index, a biomarker, often used to evaluate severity of NAFLD, may be useful to evaluate risk for incident DM in ordinary clinical setting. Here, we determined the association of FIB-4 index with changes in indices representing glucose metabolism with aging in a non-diabetic population. From among the participants of the population-based Iwaki study of Japanese people conducted during 2014-2017, 1,268 non-diabetic individuals with complete data sets (age: 51.4 ± 15.9 years; men/women: 485/773) were enrolled in a cross-sectional study. In addition, of the participants, 439 who attended consecutive appointments between 2014 and 2017 were enrolled in a longitudinal study that aimed to evaluate the changes in insulin secretion and resistance with aging (age: 53.1 ± 13.7 years; men/women: 178/261). The cross-sectional study showed significant correlations of FIB-4 index with homeostasis model of assessment (HOMA) indices, even after adjustment for multiple factors (HOMA-β: β = - 0.254, p < 0.001; HOMA-R: β = - 0.247, p < 0.001). The longitudinal study showed a significant association between FIB-4 index and the change in HOMA-β (p < 0.001) but not HOMA-R (p = 0.639) during the 3-year study period. Use of the optimal cut-off value of the FIB-4 index for the prediction of decreased insulin secretion (HOMA-β < 30), determined using receiver operating characteristic analysis (1.592), showed that individuals at risk had a hazard ratio of 2.22 (confidence interval 1.17-4.06) for decreased insulin secretion, after adjustment for confounders. FIB-4 index may represent a useful predictor of a subsequent decrease in insulin secretion, at least in a non-diabetic Japanese population.