The thermal and erosional history of convergent plate boundaries is important for understanding the links between subduction, arc magmatism, genesis of ore deposits, topography and climate of orogenic belts. Unlike the continent-continent collision that formed many of the largest orogenic belts known today, the Central Andes of South America is a unique case where an oceanic-continent collision has given rise to the Earth's longest and second tallest orogenic belt. Over the last thirty years a plethora of models have been suggested in an attempt to explain how a plateau-type orogen formed at the leading edge of western South America. In the Central Andes most research have focussed attention on the study of the evolution of the arc and backarc, since continuous subduction erosion of the forearc has left little trace of the interplate dynamics that initiated the orogenic belt. In this article, we present a new insight into the thermal and exhumation history of the forearc along the Coastal Cordillera of northern Chile based on biotite K-Ar, apatite fission-track, and apatite/zircon (U-Th)/He dating. We collected diorite samples in a 2 km thick crustal section at the coastal cliff (~. 22°S), and a sea level isoelevation profile between 21 and 27°S. Results from all three dating methods show that the cooling of Coastal Cordillera took place shortly after emplacement during a period of rifting in Jurassic times. Cooling took place in two episodes, mainly in Late Jurassic-Early Cretaceous (~ 118-152 Ma) but also during Late Cretaceous (60-80 Ma) due to the resumption of compression, rift closure, arc uplift, exhumation, eastward migration of magmatic arc activity, and thermal relaxation. The youngest apatite (U-Th)/He ages reveal a cooling event, never reported previously, between 40 and 50 Ma (Eocene). This thermal event affected a > 500 km long and > 1 km thick section of the Coastal Cordillera in northern Chile. Rock cooling recorded in the Eocene cannot be explained by the thermal effect of the magmatic arc. We explore two scenarios that can provide an explanation for the observed cooling; 1) forearc uplift and exhumation, and 2) changes in plate subduction dynamics. Erosion rates of 0.24 to 0.36 km/Myr, for a period of 10 Myr, are necessary to explain the cooling event. Alternatively, the subduction of a ridge could explain significant cooling by flat slab dewatering. Based on recent sea floor spreading reconstruction, we suggest the subduction of the Farallon-Phoenix ridge as a possible candidate. The subduction of this ridge can account for the cooling event at 40-50 Ma, but also for some important aspects of the Eocene development of the Central Andes; notably the onset of uplift, an abrupt decrease in magmatic activity, and porphyry copper mineralization. © 2010 Elsevier B.V.