BACKGROUND: The low FODMAP diet (LFD) leads to clinical response in 50%-80% of patients with irritable bowel syndrome (IBS). It is unclear why only some patients respond. AIMS: To determine if differences in baseline faecal microbiota or faecal and urine metabolite profiles may separate clinical responders to the diet from non-responders allowing predictive algorithms to be proposed. METHODS: We recruited adults fulfilling Rome III criteria for IBS to a blinded randomised controlled trial. Patients were randomised to sham diet with a placebo supplement (control) or LFD supplemented with either placebo (LFD) or 1.8 g/d B-galactooligosaccharide (LFD/B-GOS), for 4 weeks. Clinical response was defined as adequate symptom relief at 4 weeks after the intervention (global symptom question). Differences between responders and non-responders in faecal microbiota (FISH, 16S rRNA sequencing) and faecal (gas-liquid chromatography, gas-chromatography mass-spectrometry) and urine ( H NMR) metabolites were analysed. RESULTS: At 4 weeks, clinical response differed across the 3groups with adequate symptom relief of 30% (7/23) in controls, 50% (11/22) in the LFD group and 67% (16/24) in the LFD/B-GOS group (p = 0.048). In the control and the LFD/B-GOS groups, microbiota and metabolites did not separate responders from non-responders. In the LFD group, higher baseline faecal propionate (sensitivity 91%, specificity 89%) and cyclohexanecarboxylic acid esters (sensitivity 80%, specificity 78%), and urine metabolite profile (Q 0.296 vs. randomised -0.175) predicted clinical response. CONCLUSIONS: Baseline faecal and urine metabolites may predict response to the LFD.