The aims of this work were to study on dairy farm conditions: i) the repeatability of long-term enteric CH emissions measurement from lactating dairy cows using GreenFeed (GF); ii) the ranking of dairy cows according to their CH emissions across diets. Forty-five Holstein lactating dairy cows were randomly assigned to 3 equivalent groups at the beginning of their lactation. The experiment was composed of 3 successive periods: i) pre-experimental period (weeks 1 to 5) in which all cows received a common diet; ii) a dietary treatment transition period (weeks 6 to 10); and iii) an experimental period (weeks 11 to 26) in which each group was fed a different diet. Experimental diets were formulated to generate more or less CH production: i) a diet based on ryegrass silage and concentrates, low in starch and lipid, designed to induce high CH emissions (CH4+); ii) a diet based on maize silage and concentrates, rich in starch, designed to induce intermediate CH emissions (CH4int); iii) a diet based on maize silage and concentrates, rich in starch and lipid, designed to induce low CH emissions (CH4-). Gas emissions were individually measured using GF systems. Repeatability of gas emissions, dry matter intake (DMI) and dairy performances measurements was calculated from data averaged over 1, 2, 4, and 8 weeks for each animal. Hierarchical cluster analysis was performed to rank individual animals according to their CH emissions. No significant differences were observed for daily CH emissions (g/day) among diets, because of lower DMI of CH4+ cows. When CH emissions were referred to units of DMI or milk, the differences among diets emerged as significant and persistent over the observed period of lactation. Repeatability values of gas emissions measurements were higher than 0.7 averaged over 8 weeks of measurement, but still higher than 0.6 for CH g/day, CO g/day, CH g/kg milk, and CH/CO even averaging only 2 weeks of measurement. The repeatability of CH emissions measurement was systematically lower than those of DMI or dairy performance parameters, like milk and FPCM yield, irrespective of the averaged measurement period. The dairy cow ranking was not stable over time between all individuals or within any of the diets. In our experimental conditions, the GF performance in the long term can be considered reliable in differentiating dairy herds by their CH emissions according to diets with different methanogenic potential, but did not allow the ranking of individual dairy cows within a same diet. Our data highlight the importance of phenotyping animals across environment in which they will be expected to perform.