Photodissociation of iodo-tyrosine modified peptides yields localized radicals on the tyrosine side chain, which can be further dissociated by collisional activation. We have performed extensive experiments on model peptides, RGYALG, RGYG, and their derivatives, to elucidate the mechanisms underlying backbone fragmentation at tyrosine. Neither acetylation nor deuteration of the tyrosyl phenolic hydrogen significantly affects backbone fragmentation. However, deuterium migration from the tyrosyl beta carbon is concomitant with cleavage at tyrosine. Substitution of tyrosine with 4-hydroxyphenylglycine, which does not have beta hydrogens, results in almost complete elimination of backbone fragmentation at tyrosine. These results suggest that a radical situated on the beta carbon is required for a-type fragmentation in hydrogen-deficient radical peptides. Replacement of the alphaH of the residue adjacent to tyrosine with methyl groups results in significant diminution of backbone fragmentation. The initial radical abstracts an alphaH from the adjacent amino acid, which is poised to "rebound" and abstract the betaH of tyrosine through a six-membered transition-state. Subsequent beta-scission leads to the observed a-type backbone fragment. These results from deuterated peptides clearly reveal that radical migration in peptides can occur and that multiple migrations are not infrequent. Counterintuitively, close examination of all experimental results reveals that the probability for fragmentation at a particular residue is well correlated with thermodynamic radical stability. A-type fragmentation therefore appears to be most likely when favorable thermodynamics are combined with the relevant kinetic control. These results are consistent with ab initio calculations, which demonstrate that barriers to migration are significantly smaller in magnitude than probable dissociation thresholds.