Given the role of the rumen microbiome in providing nutrients to the host ruminant, it is expected that rumen microbes contribute to inter-animal variations in feed efficiency. However, the link between microbial structure and an "efficient" host phenotype is unclear. We hypothesized that extreme residual feed intake (RFI) phenotypes would display distinctive microbiome features regardless of the diet. In this study, we selected the 32 most extreme RFI Charolais bulls from a cohort of 100 animals fed corn-silage (CS; n = 50) or grass-silage (GS; n = 50) based diets. Rumen samples were obtained 3 h after feeding, at slaughter, for fermentation and metataxonomic and metatranscriptomic microbial analysis. Volatile fatty acid profiles showed no differences between diets and between extreme RFI phenotypes (P > 0.05). Total bacteria and methanogen populations did not differ between extreme RFI phenotypes (P > 0.05), although methanogens expressed per liquid rumen digesta weight tended to decrease in the most efficient bulls compared to the least efficient ones (P = 0.10). The rumen microbial community structure differed between diets (P < 0.001), and between extreme RFI phenotypes in the GS diet. In the whole dataset, we identified Succiniclasticum, Saccharofermentans, Clostridia_258483 and CAG-238 as bacteria discriminant between extreme RFI phenotypes (q < 0.10). Within diets, these four genera were also discriminant in the GS diet and were all associated with the least efficient bulls. In contrast, in bulls fed the CS diet, only Saccharofermentans and Succiniclasticum were discriminant but they were positively associated with the most efficient bulls. Rumen microbial functional features did not differ between extreme RFI phenotypes but did differ between diets. In conclusion, the rumen microbiome was mainly influenced by diet, with the RFI phenotype being a marginal effector. Succiniclasticum, Saccharofermentans, Clostridia_258483, and CAG-238 were discriminant between extreme RFI phenotypes regardless of diet. However, the direction of the association with RFI was diet dependent, indicating a diet-RFI interaction and suggesting that these discriminating microbes may be suitable microbial indicator species for RFI only when considered in conjunction with the diet information.(c) 2025 The Authors. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd.