An electrical resistivity profile across the central Mariana subduction system shows high resistivity in the upper mantle beneath the back-arc spreading ridge where melt might be expected to exist. Although seismic data are equivocal on the extent of a possible melt region, the question arises as to why a 2-D magnetotelluric (MT) survey apparently failed to image any melt. We have run forward models and inversions that test possible 3-D melt geometries that are consistent with the MT data and results of other studies from the region, and that we use to place upper bounds on the possible extent of 3-D melt region beneath the spreading center. Our study suggests that the largest melt region that was not directly imaged by the 2-D MT data, but that is compatible with the observations as well as the likely effects of melt focusing, has a 3-D shape on a ridge-segment scale focused toward the spreading center and a resistivity of 100 ?-m that corresponds to ?0.1–?1% interconnected silicate melt embedded in a background resistivity of ?500 ?-m. In contrast to the superfast spreading southern East Pacific Rise, the 3-D melt region suggests that buoyant mantle upwelling on a ridge-segment scale is the dominant process beneath the slow-spreading central Mariana back-arc. A final test considers whether the inability to image a 3-D melt region was a result of the 2-D survey geometry. The result reveals that the 2-D transect completed is useful to elucidate a broad range of 3-D melt bodies.