Metabolites produced by the intestinal microbiota are potentially important physiological modulators. Here we present a metabolomics strategy that models microbiota metabolism as a reaction network and utilizes pathway analysis to facilitate identification and characterization of microbiota metabolites. Of the 2,409 reactions in the model, ~53% do not occur in the host, and thus represent functions dependent on the microbiota. The largest group of such reactions involves amino-acid metabolism. Focusing on aromatic amino acids, we predict metabolic products that can be derived from these sources, while discriminating between microbiota- and host-dependent derivatives. We confirm the presence of 26 out of 49 predicted metabolites, and quantify their levels in the caecum of control and germ-free mice using two independent mass spectrometry methods. We further investigate the bioactivity of the confirmed metabolites, and identify two microbiota-generated metabolites (5-hydroxy-L-tryptophan and salicylate) as activators of the aryl hydrocarbon receptor.