Abstract Objectives Fibroblast Growth Factor 21 (FGF21), a response to metabolic stress, is influenced by the dietary protein content. Previous studies have shown that a protein level below 10% of energy increases FGF21 hepatic secretion in mice, and increases food intake and energy expenditure. However, it has also been shown in vitro that glucose stimulates FGF21 secretion in liver. The objective of this study was to determine the respective roles of dietary protein and carbohydrate contents on FGF21 secretion and associated metabolic responses. Methods 70 male Wistar rats were subjected at one to 12 diets with various milk protein contents (3, 5, 8, 15 and 30% P of energy) and a mixture of carbohydrates and fats in which carbohydrates amounted 30, 45, 60 or 75% of energy. Body weight and energy intake were measured twice a week, and energy expenditure was measured one time after 2 weeks by indirect calorimetry. After 3 weeks, plasma was collected and an ELISA test was used to determine plasma concentration of FGF21. Tissues were dissected and weighed to determine body composition. Pieces of liver were frozen for measuring expression of Fgf21 mRNA by RT-PCR. Statistical analyzes were done by analysis of variance. Results The decrease in %P in diets increased liver Fgf21 mRNA. Using the 30% P fed group as a reference, Fgf21 mRNA were increased not significantly 3x in 15% P, but significantly 19x in 8% P, 44x in 5% P and 60x in 3% P (P < 0.001). This was related to an increase of plasma FGF21. In contrast, dietary carbohydrate contents did not affect FGF21. In response to the increased of FGF21 secretion, energy intake was increased at 8% and 5% P and was decreased in 3% P fed mice; and energy expenditure was increased in 5% and 3% P. Finally, weight gain was negative at 3% P, and lower in at 8% and 5% P than in 15% and 30% P, consequently to a lean body mass smaller in 3%, 5% and 8% P. Conclusions Liver expression of Fgf21 mRNA and plasma FGF21 increased sharply in response to the decrease in dietary protein levels confirming the role of FGF21 in signaling protein deficiency and associated metabolic and behavioral responses. The lack of effect of the carbohydrate content of diets suggests that, in vivo, only protein content affects FGF21. At last, it should be noted that despite an apparently optimal growth at 15% P, FGF21 secretion is already higher than at 30% P. Funding Sources The funding of the experiments is provided by UMR PNCA.