Equine grass sickness (EGS) is a frequently fatal disease of horses, responsible for the death of 1-2% of the UK horse population annually. The etiology of this disease is currently uncharacterized although there is evidence it is associated with Clostridium botulinum neurotoxin in the gut. Prevention is currently not possible and ileal biopsy diagnosis is invasive. The aim of this study was to characterize the fecal microbiota and biofluid metabolic profiles of EGS horses, to further understand the mechanisms underlying this disease and identify metabolic biomarkers to aid in diagnosis. Urine, plasma and feces were collected from horses with EGS, matched controls (MC), and hospital controls (HC). Sequencing the16S rRNA gene of the fecal bacterial population of the study horses found a severe dysbiosis in EGS horses, with an increase in Bacteroidetes and a decrease in Firmicutes bacteria. Metabolic profiling by 1H nuclear magnetic resonance (NMR) spectroscopy found EGS to be associated with the lower urinary excretion of hippurate and 4-cresyl sulfate and higher excretion of O-acetyl carnitine and trimethylamine-N-oxide (TMAO). The predictive ability of the complete urinary metabolic signature and using the four discriminatory urinary metabolites to classify horses by disease status was assessed using a second (test) set of horses. The urinary metabolome and a combination of the four candidate biomarkers showed promise in aiding the identification of horses with EGS. Characterization of the metabolic shifts associated with EGS offers the potential of a non-invasive test to aid pre-mortem diagnosis.