Extensive sampling, major element chemistry on over 300 samples and K-Ar radiometric dating have been carried out on the Ecuadorian Upper Tertiary–Quaternary volcanoes. The results show important space–time variations of the volcanic activity, between Late Miocene time and the present. In Late Miocene time a continuous volcanic belt, located approximately along the present volcanic front (VF), affected the whole country from the Cuenca basin to the south, up to Colombia to the north. Major changes occurred at about 5 Ma: volcanic activity stopped south of the Guayaquil fault belt and never resumed; to the north the active volcanic axis shifted eastward to the Cordillera Real (CR) area with a simultaneous relative decrease in intensity. Since Early Quaternary time the volcanic belt widened westward to the Western Cordillera where the volcanism resumed at about 1.5–1.0 Ma, giving rise to the very wide active volcanic zone of Ecuador. The Plio-Quaternary products show significant longitudinal and latitudinal chemical and mineralogical changes. Volcanics of the VF and Interandean Depression contain amphibole and define a calc-alkaline trend with a K 2 O content lower than that of the CR products, which are characterized by a mostly anhydrous phenocryst assemblage. In both areas andesites dominate, but extreme compositions (basaltic andesites and rhyolites) are more diffuse in the CR than the VF. No significant transverse zoning has been detected in the northern region (north of the Chota-Mira transverse tectonic line). The observed temporal and spatial variations are interpreted as a result of the subduction of the Carnegie Ridge anomalous oceanic crust, underthrusting of which began approximately 6 Ma ago.