Protein stability is enhanced by the addition of osmolytes, such as sugars and polyols and inert crowders, such as polyethylene glycols. This stability enhancement has been quantified by the preferential hydration parameter which can be determined by experiments. To understand the mechanism of protein stability enhancement, we present a statistical mechanical analysis of the preferential hydration parameter based upon Kirkwood-Buff theory. Previously, the preferential hydration parameter was interpreted in terms of the number of hydration waters, as well as the cosolvent exclusion volume. It was not clear how accurate these interpretations were, nor what the relationship is between the two. By using the Kirkwood-Buff theory and experimental data, we conclude that the contribution from the cosolvent exclusion dominantly determines the preferential hydration parameters for crowders. For osmolytes, although the cosolvent exclusion largely determines the preferential hydration parameters, the contribution from hydration may not be negligible.