Hydrogen bonds profoundly influence the architecture and activity of biological macromolecules. Deep appreciation of hydrogen bond contributions to biomolecular function thus requires a detailed understanding of hydrogen bond structure and energetics and the relationship between these properties. Hydrogen bond formation energies (ΔGf) are enormously more favorable in aprotic solvents than in water, and two classes of contributing factors have been proposed to explain this energetic difference, focusing respectively on the isolated and hydrogen-bonded species: (I) water stabilizes the dissociated donor and acceptor groups much better than aprotic solvents, thereby reducing the driving force for hydrogen bond formation; and (II) water lengthens hydrogen bonds compared to aprotic environments, thereby decreasing the potential energy within the hydrogen bond. Each model has been proposed to provide a dominant contribution to ΔGf, but incisive tests that distinguish the importance of these contributions are lacking. Here we directly test the structural basis of model II. Neutron crystallography, NMR spectroscopy, and quantum mechanical calculations demonstrate that O-H···O hydrogen bonds in crystals, chloroform, acetone, and water have nearly identical lengths and very similar potential energy surfaces despite ΔGf differences >8 kcal/mol across these solvents. These results rule out a substantial contribution from solvent-dependent differences in hydrogen bond structure and potential energy after association (model II) and thus support the conclusion that differences in hydrogen bond ΔGf are predominantly determined by solvent interactions with the dissociated groups (model I). These findings advance our understanding of universal hydrogen-bonding interactions and have important implications for biology and engineering.