Geochemical studies of long-lived volcanic complexes are crucial for the understanding of the nature and composition of the subduction component of arc magmatism. The Pichincha Volcanic Complex (Northern Andean Volcanic Zone) consists of: (1) an old, highly eroded edifice, the Rucu Pichincha, whose lavas are mostly andesites, erupted from 1,100 to 150 ka; and (2) a younger, essentially dacitic, Guagua Pichincha composite edifice, with three main construction phases (Basal Guagua Pichincha, Toaza, and Cristal) which developed over the last 60 ka. This structural evolution was accompanied by a progressive increase of most incompatible trace element abundances and ratios, as well as by a sharp decrease of fluid-mobile to fluid-immobile element ratios. Geochemical data indicate that fractional crystallization of an amphibole-rich cumulate may account for the evolution from the Guagua Pichincha andesites to dacites. However, in order to explain the transition between the Rucu Pichincha andesites and Guagua Pichincha dacites, the mineralogical and geochemical data indicate the predominance of magma mixing processes between a mafic, trace-element depleted, mantle-derived end-member, and a siliceous, trace-element enriched, adakitic end-member. The systematic variation of trace element abundances and ratios in primitive samples leads us to propose that the Rucu Pichincha magmas came from a hydrous-fluid metasomatized mantle wedge, whereas Guagua Pichincha magmas are related to partial melting of a siliceous-melt metasomatized mantle. This temporal evolution implies a change from dehydration to partial melting of the slab, which may be associated with an increase in the geothermal gradient along the slab due to the presence of the subducted Carnegie Ridge at the subduction system. This work emphasizes the importance of studying arc-magma systems over long periods of time (of at least 1 million of years), in order to evaluate the potential variations of the slab contribution into the mantle source of the arc magmatism.