The palladium-catalyzed allylation of primary amines has been investigated by DFT calculations (B3PW91, PCM method), and two potential mechanisms were studied. The first mechanism relies on the formation of cationic hydridopalladium complexes. Their formation involves a metal-assisted formal (1,3) shift of a proton from the nitrogen atom of an ammonium to the Cbeta carbon atom. The second part of the cycle relies on a ligand exchange through a pentacoordinated 18VE hydridopalladium complex. The last step likely proceeds through a bimolecular pathway and formally consists of a proton transfer from the allylammonium to the alcohol group of the complex. The second mechanism, which is closer to that currently admitted for nucleophilic allylic substitutions, relies on the decomplexation of the coordinated allylammonium and appears to be favored. This catalytic cycle was recomputed on model complexes varying the ligands, and a charge decomposition analysis was carried out to assess the influence of the electronic properties of the ligands. To compare our results with competitive experiments, CDA calculations were also performed on real ligands. In agreement with experimental observations, this process was found to be strongly ligand dependent, decomplexation being favored by strong pi-acceptor ligands. These calculations led us to show experimentally that complex [Pd(P(OPh)(3))(2)(eta(3)-C(3)H(5))][OTf] is an efficient catalyst for this allylation. Finally, this catalytic process proved to be sensitive to the nature of the amine, with poorly basic amines favoring the re-formation of the catalytic precursor.