Abstract. It is well-known that oxazolone b 2 ions fragment extensively by elimination of CO to form a 2 ions, which often fragment further to form a 1 ions. Less well-known is that some oxazolone b 2 ions may fragment directly to form a 1 ions. The present study uses energy-resolved collision-induced dissociation experiments to explore the occurrence of the direct b 2 →a 1 fragmentation reaction. The experimental results show that the direct b 2 →a 1 reaction is generally observed when Gly is the C-terminal residue of the oxazolone. When the C-terminal residue is more complex, it is able to provide increased stability of the a 2 product in the b 2 →a 2 fragmentation pathway. Our computational studies of the relative critical reaction energies for the b 2 →a 2 reaction compared with those for the b 2 →a 1 reaction provide support that the critical reaction energies are similar for the two pathways when the C-terminal residue of the oxazolone is Gly. By contrast, when the nitrogen of the oxazolone ring in the b 2 ion does not bear a hydrogen, as in the Ala-Sar and Tyr-Sar (Sar=N-methylglycine) oxazolone b 2 ions, a 1 ions are not formed but rather neutral imine elimination from the N-terminus of the b 2 ion becomes a dominant fragmentation reaction. The M06-2X/6-31+G(d,p) density functional theory calculations are in general agreement with the experimental data for both types of reaction. In contrast, the B3LYP/6-31+G(d,p) model systematically underestimates the barriers of these S N 2-like b 2 →a 1 reaction. The difference between the two methods of barrier calculation are highly significant (PG0.001) for the b 2 →a 1 reaction, but only marginally significant (P=0.05) for the b 2 →a 2 reaction. The computations provide further evidence of the limitations of the B3LYP functional when describing S N 2-like reactions.