Whole-rock geochemical data on basaltic to rhyolitic samples from 12 late Pleistocene-Holocene centers are used to constrain the role of continental crust in the genesis of melts formed beneath the anomalously wide arc in Ecuador. Primitive and relatively homogeneous isotopic compositions observed across the arc imply that Ecuador lavas were produced largely within the subduction zone and not by extensive melting of crustal rocks similar to those upon-which the volcanoes were built. Mixing calculations limit the quantity of assimilated crust to less than approximately 10%. Cross-arc geochemical variation (e.g., in 143Nd/144Nd, Ba/Nb, La/Yb) indicates that assimilation of crustal rock is most active in the center-arc area, around the Inter-Andean Graben, at distances from 330-360 km from the trench. Characteristically 'adakitic' features (low Y, Yb, high La/Yb), which are observed in virtually all andesites and dacites in Ecuador as well as throughout the central and southern Andes, appear to be strongly influenced by crystal fractionation (e.g., as indicated by compatible element behavior for Y) and are most clearly developed in lavas that, based on isotopic compositions, have assimilated the most crust. A subset of andesites, which display an unusual combination of high Sr (>900 ppm) and non-radiogenic isotopes (eNd>4.1, D7/4Pb<6.0), appear to be regionally distinctive in Ecuador (i.e., not observed among modern lavas in the Andean central and southern volcanic zones or in Colombia). Only these lavas, which are well represented at Imbabura Volcano, appear to have geochemical features that may be consistent with a slab-melt interpretation associated with the subduction of relatively young, over-thickened oceanic crust of the subducting Carnegie Ridge.