The pH dependence of oxygen affinity of hemoglobin (Bohr effect) is due to ligand-linked pK shifts of ionizable groups. Attempt to identify these groups has produced controversial data and interpretations. In a further attempt to clarify the situation, we noticed that hemoglobin alkylated in its liganded form lost the Bohr effect while hemoglobin alkylated in its unliganded form showed the presence of a practically unmodified Bohr effect. In spite of this difference, analyses of the extent of alkylation of the two compounds failed to identify the presence of specific preferential alkylations. In particular, the alpha 1 valines and beta 146 histidines appeared to be alkylated to the same extent in the two proteins. Focusing our attention on the effect of the anions on the functional properties of hemoglobin, we measured the Bohr effect of untreated hemoglobin in buffers made with HEPES [N-(2-hydroxyethyl)piperazine-N'-2-ethanesulfonic acid], MES [2-(N-morpholino)ethanesulfonic acid], and MOPS [3-(N-morpholino)propanesulfonic acid], which being zwitterions do not need addition of chlorides or other anions for reaching the desired pH. The shape acquired by the Bohr effect curves, either as pH dependence of oxygen affinity or as pH dependence of protons exchanged with the solution, was irreconcilable with that of the Bohr effect curves in usual buffers. This indicated the relevance of solvent components in determining the functional properties of hemoglobin. A new thermodynamic model is proposed for the Bohr effect that includes the interaction of hemoglobin with solvent components. The classic proton Bohr effect is a special case of the new theory.