The long-term effect of a plant (P)-based diet was assessed by proton nuclear magnetic resonance (1 H-NMR) metabolomics in rainbow trout fed a marine fish meal (FM)-fish oil (FO) diet (M), a P-based diet and a control commercial-like diet (C) starting with the first feeding. Growth performances were not heavily altered by long-term feeding on the P-based diet. An 1 H-NMR metabolomic analysis of the feed revealed significantly different soluble chemical compound profiles between the diets. A set of soluble chemical compounds was found to be specific either to the P-based diet or to the M diet. Pterin, a biomarker of plant feedstuffs, was identified both in the P-based diet and in the plasma of fish fed the P-based diet. 1 H-NMR metabolomic analysis on fish plasma and liver and muscle tissues at 6 and 48 h post feeding revealed significantly different profiles between the P-based diet and the M diet, while the C diet showed intermediate results. A higher amino acid content was found in the plasma of fish fed the P-based diet compared with the M diet after 48 h, suggesting either a delayed delivery of the amino acids or a lower amino acid utilisation in the P-based diet. This was associated with an accumulation of essential amino acids and the depletion of glutamine in the muscle, together with an accumulation of choline in the liver. Combined with an anticipated absorption of methionine and lysine supplemented in free form, the present results suggest an imbalanced essential amino acid supply for protein metabolism in the muscle and for specific functions of the liver.