The Mesozoic magnetite-(apatite) deposits of the Coastal Cordillera of Chile are interpreted as the product of the crystallization of oxidized iron-rich melts and subsequent hydrothermal alteration produced by related magmatic-hydrothermal systems. These deposits form a regional-scale mineral system controlled by the Atacama Fault System and where the mineralization spans more than 10 km in vertical extent. Individual sub-vertical bodies of massive magnetite coexist with and evolve vertically into pegmatite-, breccia-, and vein-like apatite-actinolite-magnetite/ilmenite rock. The mineralization is always hosted by a hydrothermal aureole of alkali-calcic-iron alteration that includes stockwork-like to disseminated mineralization. The deposits cluster in two groups. Those located in the northern part are mostly vein-like, and are hosted by Jurassic diorite. They have(87)Sr/(86)Sr(i)and epsilon Nd(i)values of 0.7042-0.7062 and + 5.1 to + 7.2, respectively. The southern group includes shallowly emplaced ore lenses in broadly coeval (sub-)volcanic intermediate rocks. They show similar(87)Sr/(86)Sr(i)signatures (0.7033-0.7065, with one value up to 0.7097) and more variable epsilon Nd(i)values (+ 3.9 to + 8.6). As a whole, the Sr-Nd data do not seem to be influenced by the type of crust intruded, but rather, likely track the mixing between a MORB-like reservoir and another source with elevated(87)Sr/Sr-86(i)(>= 0.706). The genetic model proposed involves the dehydration of variably altered subducted oceanic crust, the interaction of fluids released from the mantle wedge, the separation of iron-rich melts, and their ascent along transcrustal faults. The broadly coeval intermediate host rocks show a lesser contribution of subducted crust, something that perhaps excludes a genetic relationship between these rocks and the magnetite-(apatite) mineralization.