Background. In clinical practice, there is a strong interest in non-invasive markers of non-alcoholic fatty liver disease (NAFLD). Our hypothesis was that the fold-change in plasma triglycerides (TG) during a 2-h oral glucose tolerance test (fold-change TG(OGTT)) in concert with blood glucose and lipid parameters, and the rs738409 C > G single nucleotide polymorphism (SNP) in PNPLA3 might improve the power of the widely used fatty liver index (FLI) to predict NAFLD. Methods. The liver fat content of 330 subjects was quantified by H-1-magnetic resonance spectroscopy. Blood parameters were measured during fasting and after a 2-h OGTT. A subgroup of 213 subjects underwent these measurements before and after 9 months of a lifestyle intervention. Results. The fold-change TG(OGTT) was closely associated with liver fat content (r=0.51, P < 0.0001), but had less power to predict NAFLD (AUROC = 0.75) than the FLI (AUROC = 0.79). Not only was the fold-change TG(OGTT) independently associated with liver fat content and NAFLD, but so also were the 2-h blood glucose level and rs738409 C>G SNP in PNPLA3. In fact, a novel index (extended FLI) generated from these and the usual FLI parameters considerably increased its power to predict NAFLD (AUROC = 0.79-0.86). The extended FLI also increased the power to predict changes in liver fat content with a lifestyle intervention (n = 213; standardized beta coefficient: 0.23-0.29). Conclusion. This study has provided novel data confirming that the OGTT-derived fold-change TG(OGTT) and 2-h glucose level, together with the rs738409 C>G SNP in PNPLA3, allow calculation of an extended FLI that considerably improves its power to predict NAFLD. (C) 2017 Elsevier Masson SAS. All rights reserved.