The gut microbiota helps us digest food and has been associated with both good health and risk of disease. Although compositions of the gut microbiota are highly divergent, the gut microbiota exhibits a high level of functional redundancy making it difficult to reliably link gut microbiome patterns to health and diet across individuals. Thus, there is a need to move beyond profiling of the gut microbial composition to the assessment of gut microbial metabolic activity in order to expand knowledge on the impact of the gut microbiota on health. Metabolomics has already proven its utility in identifying gut-derived microbial metabolites, which may be mediators of diet-induced host–microbial crosstalk important for health. These diet-derived metabolites include, among many others, the short-chain fatty acids originating from bacterial degradation of dietary fibres and proteins, secondary bile acids derived from primary bile acids, microbial tryptophan catabolites resulting from proteolysis, imidazole propionate produced from histidine and trimethylamine N-oxide, a host–microbial co-metabolite of nutrients such as phosphatidylcholine, choline and L-carnitine, present in high-fat foods. These co-metabolites have been associated with beneficial and detrimental effects in the host. However, the vast majority of metabolites measured by untargeted metabolomics in human blood and excreta remain unknown and may include additional microbial molecules of importance for health. Thus, there is great potential, yet to be explored, for identifying additional diet-derived microbial products that affect the host. Herein, we review key advances, challenges and limitations in our understanding of diet–microbiome interactions and diet-derived microbial metabolites in relation to human health and discuss how metabolomics may shed light on inter-individual differences in dietary responses.