The current rule to predict the enantiopreference of subtilisin toward secondary alcohols is based on the size of the substituents at the stereocenter and implies that the active site contains two differently sized pockets for these substituents. Several experiments are inconsistent with the current rule. First, the X-ray structures of subtilisin show there is only one pocket (the S1- pocket) approximately the size of a phenyl group to bind secondary alcohols. Second, the rule often predicts the incorrect enantiomer for reactions in water. To resolve these contradictions, we refine the current rule to show that subtilisin binds only one substituent of a secondary alcohol and leaves the other in solvent. To test this refined empirical rule, we show that the enantioselectivity of a series of secondary alcohols in water varied linearly with the difference in hydrophobicity (log P/P0) of the substituents. This hydrophobicity difference accounts for the solvation of one substituent in water.