Of primary interest for every enzyme is the identification of its physiological substrates. However, the vast structural diversity of endogenous metabolites, coupled with the overlapping activities of numerous enzymes, makes it difficult to deduce the identity of natural substrates for a given enzyme based on in vitro experiments. To address this challenge, we recently introduced an LC-MS based analytical method termed discovery metabolite profiling (DMP) to evaluate the global metabolic effects of enzyme inactivation in vivo. We have applied DMP to study mice lacking the enzyme fatty acid amide hydrolase (FAAH), which degrades the endocannabinoid family of signaling lipids. DMP identified several previously uncharacterized FAAH substrates, including a structurally novel class of brain lipids that represent conjugates of very long chain fatty acids with the amino acid derivative taurine [N-acyl taurines (NATs)]. These findings show that DMP can establish direct connections between the proteome and metabolome and thus offers a powerful strategy to assign physiological functions to enzymes in the post-genomic era.