In ruminants, feed efficiency is a complex trait involving the host and its ruminal microbiota. In meat sheep, little is known about their joint action on animal efficiency. The objective of this work was to explore the ruminal and plasmatic metabolome differences between genetically divergent lines on residual feed intake (RFI). Rumen and blood samples were taken from male lambs belonging to the two divergent lines selected on RFI estimated under a 100% concentrate diet. Samples were taken twice: firstly when lambs were around 3 month-old and fed the same diet as their ancestors did during selection and secondly when lambs were 7-9 month-old and fed a diet with 1/3 concentrate and 2/3 dactylis hay. Rumen samples were taken using an oesophageal probe and blood samples were taken at the jugular vein before being centrifuged to separate plasma from blood cells. All the samples were frozen. Preparation of the samples consisted in several centrifugations and addition of a phosphate buffer. Samples were then analysed using NMR spectrometry and integrated with the ASICS R package. Data was then analysed both using a mixed linear model using the sasLM R package and multivariate analysis using mixOmics R package (PLS and PLS-DA). Because of the great effect of diet, concentrate and forage based data must be separately analysed. Rumen samples did not present many differences between lines under the concentrate diet and under the forage diet, no metabolite was differently expressed. Multivariate analysis on rumen samples did not separate the two lines well. On the contrary, plasma samples showed interesting differences between the 2 lines with amino acids being (under both diets) more present in plasma samples from less efficient animals than in samples from efficient ones. In both diets, plasma samples from efficient lambs had significantly (P<0.05) more citrate than samples from less efficient lambs. It seems that plasmatic metabolome gives a better insight in the understanding of the differences between lines, ruminal metabolome will be analysed with microbiota data in ulterior works