This paper quantifies the flexural subsidence expected from loading by a volcanic arc. The resulting mathematical model shows that the arc width should grow with time and that the subsidence beneath the load can be estimated from the observed arc width at the surface. Application of this model to the Halmahera Arc in Indonesia shows an excellent fit to observations if a broken-plate model of flexure is assumed. The model also gives an excellent fit to data from East Java, also in Indonesia, where it is possible to forward model gravity anomalies. In particular, the depth, location, and width of the depocenter-associated gravity low are accurately reproduced, although the model does require a high density for the volcanic arc (2900 kg m⁻³). This may indicate additional buried loads due, for example, to magmatic underplating. Our main conclusion is that loads generated by the volcanic arc are sufficient to account for much, if not all, of the subsidence in basins within ∼100 km of active volcanoes at subduction plate boundaries, if the plate is broken. The basins will be asymmetrical and, close to the arc, will contain coarse volcaniclastic material, whereas deposits farther away are likely to be volcaniclastic turbidites. The density contrast between arc and underlying crust required to produce the Indonesian arc basins means that they are unlikely to form in young intraoceanic arcs but may be common in older and more mature arcs.