Satellite gravity, ship track bathymetry, sediment thickness, and crustal magnetic age data were combined to calculate the residual bathymetry and residual mantle Bouguer gravity anomaly (RMBA) for the northwest Pacific Ocean. The Hawaii-Emperor hot spot track appears on the RMBA map as a chain of negative anomalies, implying thickened crust or less dense mantle. The hot spot swell is clearly visible in a broad band of half-width ?500 km for about 2000 km downstream from the current hot spot location, corresponding to hot spot ages of 0–25 Ma. A much narrower expression of the hot spot is visible for the rest of the chain at hot spot ages of 25–80 Ma. Comparison of the observed RMBA with various compensation models reveals that the relatively narrow features of the Hawaii-Emperor seamounts are best explained as being supported by plate flexure, while the Shatsky Rise, Hess Rise, and Mid-Pacific Mountains oceanic plateaus are best fit by Airy isostasy with a thickened crustal root. Amplitude comparisons between the RMBA predictions of various compensation models and the observed RMBA for the Hawaiian swell are ambiguous. However, on the basis of the shape of the predicted anomalies, we favor a model of flexure in response to a buried load at 120 km depth. We further calculate igneous (i.e., crustal) volume flux along the axis of the Hawaii-Emperor hot spot by integrating cross-sectional areas of gravity-derived excess crustal thickness and seafloor elevation, respectively, with respect to the normal oceanic crust. The highest values of the calculated igneous volume flux along the Hawaiian and Emperor ridges (?8 m3/s) occur at present and at about 20 Ma. The flux was reduced to only 50% of this maximum (?4 m3/s) at 10 Ma. The calculated igneous volume flux is systematically smaller (maximum values of ?4 m3/s) along the Emperor ridge. Overall, the Hawaiian and Emperor ridges appear to have experienced quasi-periodic variations in fluxes on timescales of 6–30 Ma. Furthermore, during the low-flux periods at 25–48, 57, and 75 Ma the height and size of individual hot spot seamounts appear to be noticeably less than those of the high-flux periods. We hypothesize that the variations in the fluxes of the Hawaiian ridge might be controlled by the thickness of the overlying lithosphere at the time of hot spot emplacement, while the variations along the Emperor ridge may be influenced by the dynamics of the slow absolute motion of the hot spot at the time.