Members of microbial communities interact via a plethora of mechanisms, including resource competition, cross-feeding, and pH modulation. However, the relative contributions of these mechanisms to community dynamics remain uncharacterized. Here, we develop a framework to distinguish the effects of resource competition from other interaction mechanisms by integrating data from growth measurements in spent media, synthetic community assembly, and metabolomics with consumer resource models. When applied to human gut commensals, our framework revealed that resource competition alone could explain most pairwise interactions. The resource-competition landscape inferred from metabolomic profiles of individual species predicted assembly compositions, demonstrating that resource competition is a dominant driver of in vitro community assembly. Moreover, the identification and incorporation of interactions other than resource competition, including pH-mediated effects and cross-feeding, improved model predictions. Our work provides an experimental and modeling framework to characterize and quantify interspecies interactions in vitro that should advance mechanistically principled engineering of microbial communities.