The maximum strength theorem is applied to determine the pressure conditions in accretionary wedges responsible for the activation of splay faults (SF). This theorem, based on the limit analysis developed in soil mechanics, selects among several modes of deformation the one prevailing for a given pressure distribution. It applies for frictional rock properties and for an arbitrary wedge geometry. In particular, the décollement is assumed here to be partitioned into two mechanically distinct regions, the internal region being part of the seismogenic zone. The considered modes of deformation include the activation of a thrust fold rooting on the décollement and outcropping anywhere on the topography. This mode, if close to the transition between the seismic and the aseismic region of the décollement is in competition with the SF which constitutes also a potential mode of deformation. The dominance of the various modes of deformation is presented in stability maps in the plane spanned by the pressure ratios of the two regions of the décollement (pore pressure divided by lithostatic pressure, both referenced to the sea floor). Several maps are proposed for the particular case of the Kumano transect of Nankai wedge, South-West Japan, varying the SF pressure ratio, the position of the SF root on the décollement with respect to the transition between the seismic and aseismic regions as well as the respective length of these two regions. In addition, these stability maps are suggesting various scenarios to explain the changes through time in the prevailing mode of deformation in an accretionary wedges. For the Kumano transect, three scenarios could explain the observed deactivation of the SF, one of them being the seaward migration of the décollement transition point.