Protein classification typically uses structural, sequence or functional similarity. Here we introduce an orthogonal method that organizes proteins by ligand similarity, focusing on the class A G-protein-coupled receptor (GPCR) protein family. Comparing a ligand-based dendrogram to a sequence-based one, we identified GPCRs that were distantly linked by sequence but were neighbors by ligand similarity. Experimental testing of the ligands predicted to link three of these new pairs confirmed the predicted association, with potencies ranging from low nanomolar to low micromolar. We also predicted hundreds of non-GPCRs closely related to GPCRs by ligand similarity and confirmed several cases experimentally. Ligand similarities among these targets may reflect the conservation of identical ligands among unrelated receptors, which signal in different time domains. Our method integrates these apparently disparate receptors into chemically coherent circuits and suggests which of these receptors may be targeted by individual ligands.