Abstract Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) is a complex, multi-system, debilitating disability manifesting as severe fatigue and post-exertional malaise. The chronic dysfunctions in ME/CFS are increasingly recognized as significant health factors with potential parallels with ‘long COVID’. However, the etiology of ME/CFS remains elusive with limited high-resolution human studies. In addition, reliable biomarker-based diagnostics have not been well-established, but may assist in disease classification, particularly during different temporal phases of the disease. Here, we performed deep multi-‘omics (shotgun metagenomics of gut microbiota and plasma metabolomics) and clinical phenotyping of healthy controls (n=79) vs. two cohorts of ME/CFS patients – those with short-term disease (<4 years, n=75), and patients with long-term disease (>10y, n=79). Overall, ME/CFS was characterized by reduced gut microbiome diversity and richness with high heterogeneity, and depletion of sphingomyelins and short-chain fatty acids in the plasma. We found significant differences when stratifying by cohort; short-term ME/CFS was associated with more microbial dysbiosis, but long-term ME/CFS was associated with markedly more severe phenotypic and metabolic abnormalities. We identified a reduction in the gene-coding capacity (and relative abundance of butyrate producers) of microbial butyrate biosynthesis together with a reduction in the plasma concentration of butyrate, especially in the short-term group. Global co-association and detailed gene pathway correlation analyses linking the microbiome and metabolome identified additional potential biological mechanisms underlying host-microbiome interactions in ME/CFS, including bile acids and benzoate pathways. Finally, we built multiple state-of-the-art classifiers to identify microbes, microbial gene pathways, metabolites, and clinical features that individually or together, were most able to differentiate short or long-term MECFS, or MECFS vs. healthy controls. Taken together, our study presents the highest resolution, multi-cohort and multi-‘omics analysis to date, providing an important resource to facilitate mechanistic hypotheses of host-microbiome interactions in ME/CFS.