We present quantitative models for the thermal structure of the crust and uppermost mantle in island arc areas and consider the implications of this structure for paired metamorphic belts and crustal extension behind arcs. In an active island arc (e.g., Japan), the following processes are assumed to occur: (1) The oceanic lithosphere slides under the island arc structure along the line of the oceanic trench. The uneonsolidated oceanic sediment that overlies the oceanic lithosphere accumulates against the front of the arc, building it out at a rate of about 1 km/m.y. (2) Heat is generated along the slip zone between the descending lithosphere and the rocks that overlie it. At some depth, melting occurs along the slip zone, magmas are formed, and heat is transferred upward by magmatic convection. At the surface, three distinct thermal zones run parallel to the arc; they are, in order away from the trench toward the arc: (A) zone of low heat flow, where accumulation of unconsolidated ocean floor sediment takes place; (B) zone of average heat flow occupied by previously accumulated sediment or old crust; and (C) zone of high heat flow that overlies that part of the slip zone on which melting occurs and upward transfer of heat takes place by penetrative magmatic convection. The width of each zone is variable and depends largely on the relative rate of movement of the descending lithosphere. Each zone is characterized by its own vertical geothermal gradient, and there is a strong horizontal temperature gradient between zones. At high rates of movement on the slip zone, the deeper part of zone A can provide the P/T conditions appropriate for the formation of glaucophane schists and eclogites, while zone C can give the P/T conditions for andalusite-sillimanite and kyanite-sillimanite metamorphic facies series. Upward and possibly horizontal tectonic transport of the glaucophane schists to the surface is necessary to preserve the high-pressure, low-temperature mineralogy and to give the arrangement of paired metamorphic belts observed around the margins of the Pacific. These vertical movements are probably associated with a deceleration or cessation of movement on the slip zone. The rate of magmatic intrusion into the upper crust required to give the high heat flow in zone C would result in crustal extension (spreading) behind the arc at rates of up to several centimeters, per year in a steady-state situation.