▶ Forearc gravity and magnetic anomalies are from the same causative sources. ▶ Magnetic inversion puts independent constraints on subduction zone temperature. ▶ Downdip limits of Nankai seismogenic zone do not extend down to the island arc Moho. ▶ Magnetic boundaries are likely places for slab tears triggering seismic segmentation. ▶ Ages, dip angles, and geothermal states of slab influence forearc serpentinization. I investigate large-scale deep crustal structures of the Nankai subduction zone and neighboring region using regional magnetic and gravity anomalies, heat flow measurements, and earthquake hypocenters. It is found that ages, dip angles, and geothermal states of the subducting slab have direct influences on mantle wedge serpentinization. The weakest serpentinization observed in the Nankai forearc region is associated with the youngest downgoing plate of the Shikoku Basin. Conspicuous gravity anomalies identified in the forearc region are coincidental spatially with magnetic anomalies after the reduction to the pole, a mathematical procedure that helps relocate magnetic sources and boundaries, and allows us to more easily interpret magnetic data. It is argued that these patches of magnetic and gravity anomalies are caused by the same sources of anomalous density and magnetization, and are linked directly to preexisting structures such as magnetic anomalies and their boundaries in the subducting oceanic crust. Since the gravity and magnetic anomaly patches are found to be closely related to interplate seismogenic behaviors in the Nankai subduction zone, I suggest that major magnetic boundaries in the Shikoku Basin are likely weak places for slab tears that trigger seismic segmentations along the subduction zone. Application of the Parker–Oldenburg algorithm to Bouguer gravity anomalies yields a 3D Moho topography. Curie-point depths are also estimated from the magnetic anomalies with reduction to the pole using a windowed wavenumber-domain algorithm. Window sizes are found to have little effects on the average Curie-point depths other than lowering lateral resolutions. A wide zone of deep Curie depths is identified in southwest Japan, relating to strong influence from the subduction of the relatively young and warm Shikoku Basin crust of the Philippine Sea plate. Curie depths so obtained can be correlated well with heat flow measurements, which cluster around a theoretical curve when the average thermal conductivity is about 3.0 W/(m °C). Using constraints from both Curie depths and heat flow, I also model the shallow geothermal field of the subduction zone. Earthquake hypocenters plotted against Moho and Curie depths and geothermal fields on three transects confirm early studies that downdip limits of seismogenic zones along the Nankai plate boundary do not extend down to the island arc Moho and their temperatures are more or less close to 350 °C. Geothermal field has direct influences on earthquake distributions in the overriding island arc and accretionary prism, within the subducting oceanic lithosphere, and along the interplate boundary.