Fatty liver disease is prevalent during parturition in dairy cattle. Therefore, there is an urgent need to develop novel, sensitive biomarkers for the early diagnosis of the metabolic disorders. Macroproteomics revealed that the fecal microbial community changes significantly when animal develops fatty liver disease. The microbial changes in cows with severe fatty liver (SFL) were greater than cows with moderate fatty liver (MFL) and normal condition (Norm). This suggests that microorganisms play an important role in the pathogenesis of metabolic disorders. In this study, feces-sourced microorganisms and microbial proteins were identified and testified as novel biomarkers for the early diagnosis of fatty liver disease in cattle. For example, the AUC (area under curve) values, based on Receiver Operating Characteristics analysis, of using the combination of Lachnoanaerobaculum and Bifidobacterium (at the genus level) to discriminate MFL and SFL animals reached 0.944 and 0.867, respectively. and 0.922 and 0.985, respectively, for the combination of Bifidobacterium pseudolongum and Lachnospiraceae bacterium (at the species level). Interestingly, the differentially expressed microbial proteins are closely related to the identified microorganisms. For example, the majority of the top 20 microbial proteins with significant expression differences were derived from Bifidobacterium pseudolongum. Bifidobacterium pseudolongum was considered a prominent potential biomarker for the diagnosis of metabolic disorders, especially in fatty liver cattle. The results of this study confirm that microbial signatures have a causal contribution to the pathophysiological mechanism of non-alcoholic fatty liver disease (NAFLD), but also shed light on fecal microbiota transfer (FMT) experiments in treating NAFLD.