Transition state stabilization is essential for rate acceleration of enzymatic reactions. Despite extensive studies on various transition state structures of enzymes, an intriguing puzzle is whether an enzyme can accommodate multiple transition states (TSs) to catalyze a chemical reaction. It is experimentally challenging to study this proposition in terms of the choices of suitable enzymes and the feasibility to distinguish multiple TSs. As a paradigm with the protein lysine methyltransferase (PKMT) SET7/9 paired with its physiological substrates H3 and p53, their TSs were solved with experimental kinetic isotope effects as computational constraints. Remarkably, SET7/9 adopts two structurally distinct TSs, a nearly symmetric SN2 and an extremely early SN2, for H3K4 and p53K372 methylation, respectively. The two TSs are also different from those previously revealed for other PKMTs. The setting of multiple TSs is expected to be essential for SET7/9 and likely other PKMTs to act on broad substrates with high efficiency.