Phenotyping animal feed efficiency, the animal’s ability to transform feed into food, is challenging because it is time and labor consuming and not always feasible in livestock systems based on pasture. Thus, biomarkers should be developed and validated for a high-throughput phenotyping of feed efficiency in practical conditions. The natural 15N enrichment of animal proteins over the consumed diet (Δ15N = δ15Nanimal- δ15Ndiet) has recently been proposed as a biomarker of animal feed efficiency. This study aimed to confirm by meta-analysis the potential of Δ15N to capture the between-animal variation in feed conversion efficiency (FCE; body weight gain/ dry matter intake) in young beef cattle reared in different European conditions. For this, individual data of Δ15N measured in plasma and FCE of 468 growing-fattening bulls of different pure and cross continental breeds, from 25 different diets, 8 experiments and 3 countries (France, UK and Switzerland) were evaluated by regression analysis. All animals were tested for at least 60 days for FCE and their blood (7 experiments) or muscle tissue (1 experiment) sampled at the end of the test period. Diets were sampled throughout the feed efficiency test. Diets and animal proteins (plasma or muscle) were analysed for their natural 15N abundance (δ15N) and the Δ15N was calculated for each animal. Two models were used to assess the relationship between Δ15N and FCE at the individual level. First, a mixed regression model of FCE on Δ15N with the experiment and diet as random effects, allowing these two effects to be excluded from the explored relationship. For the second model, residuals were first obtained for FCE and Δ15N after correcting for the experiment and diet effects and then regressed on each other by simple linear regression. For the first approach, the mixed model confirmed that Δ15N is significantly (P<0.001) and negatively correlated to FCE within each diet and experiment according to the following equation: FCE=0.27 (se 0.025)-0.030 (se 0.0056) × Δ15N (RSE=0.017). For the second approach, the significant (P<0.001) correlation between both residuals confirmed the results obtained with the first approach with a similar slope: FCE=-0.031(se 0.0027) × Δ15N (RSE=0.017). Present data confirm that Δ15N can be used in future to predict between-animal variation in feed efficiency in beef cattle.