In southern Peru, the pre-Oligocene volcanic suite (the Late Cretaceous to early Paleocene silicic tuff, agglomerate, and ignimbrite of the Toquepala Group) have positive ϵ Ndi (+0.6 to +5.3) and commonly negative ϵ Sri values (-7.0 to +8.0), with 206Pb/204Pb ratios comparable to those ascribed to arc magmas. Although the former generally display unfractionated La/Yb (2 to 15) and low Th/U (0.1 to 4.2), the post-Oligocene volcanic suite, comprising early Miocene to Pleistocene calc-alkaline andesite flows and ignimbrite sheets with minor basaltic intercalations, has elevated La/Yb (9 to 34) and Th/U (3 to 8), which are correlated with negative ϵ Ndi (-0.9 to -12.0), positive ϵ Sri values (+13 to +64) and low 206Pb/204Pb. The principal trace-element and isotopic characteristics that strongly characterize the post-Oligocene volcanic suite are precisely those that confer the unusual signature to the Precambrian basement granulite and gneiss exposed in the area of study (the Arequipa Massif). The latter commonly display high Th/U (20-70) and La/Yb (14 to >57), low 206Pb/204Pb (16.11 to 17.10) but present an extreme enrichment in radiogenic Sr (ϵ Sr +396 to +999). The contrasting trace-element and isotopic signatures exhibited by both suites can be best explained if the pre- and post-Oligocene volcanic suites originated from an isotopically depleted subcontinental mantle wedge, with the ignimbrite (rhyolite) being derived by crustal fractionation from andesitic parents. In this hypothesis, the post-Oligocene magmas must have undergone substantial crustal contamination (mainly by a thick Andean crust having a large Precambrian component) during fractionation, whereas the pre-Oligocene volcanic rocks underwent little contamination, perhaps because they ascended through a thinner late Mesozoic crust. We suggest, as one possibility, that the increasing degree of crustal contamination experienced by the post-Oligocene volcanic suites is partly related to the tectonic thickening of the Precambrian basement which took place during late Mesozoic-early Cenozoic time, a process which would promote larger degrees of contamination of mantle-derived magmas upon ascent. If this assumption is correct, then the rapid Cenozoic crustal thickening of the southern Peruvian Andean crust could result from a combination of massive introduction of juvenile magmas in the lower crust, underplating of a buoyant subducted oceanic plate, and crustal shortening, the latter mechanism being perhaps similar to that actually occurring in the sub-Andes where the Brazilian Shield is underthrusting the Eastern Cordillera.