Density function theory (DFT) calculations have been carried out to investigate the binding of alcohols to the odorant binding protein LUSH from Drosophila melanogaster. LUSH is one of the few proteins known to bind to ethanol at physiologically relevant concentrations and where high-resolution structural information is available for the protein bound to alcohol at these concentrations. The structures of the LUSH–alcohol complexes identify a set of specific hydrogen-bonding interactions as critical for optimal binding of ethanol. A set of truncated models based on the structure of the LUSH–butanol complex were constructed for the wild-type and mutant (T57S, S52A, and T57A) proteins in complexes with a series of n-alcohols and for the apoprotein bound to water and for the ligand-free protein. Using both gas-phase calculations and continuum solvation model calculations, we found that the widely used DFT model, B3LYP, failed to reproduce the experimentally observed trend of increasing binding affinity with the increasing length of the alkyl chain in the alcohol. In contrast, the recently developed M05-2X DFT model successfully reproduced this subtle trend. Analysis of the results indicated that multiple factors contribute to the differences in alcohol binding affinity: the H-bonding with Thr57 and Ser52 (4–5 kcal/mol per H-bond), the desolvation contribution (4–6 kcal/mol for alcohols and 8–10 kcal/mol for water), and the other noncovalent interaction (1.2 kcal/mol per CH2 group of the alcohol alkyl chain). These results reveal the outstanding potential for using the M05-2X model in calculations of protein–substrate complexes where noncovalent interactions are important.