Gravity data across the East Pacific Rise at 9¿N and 12¿N (the ROSE area) are characterized by a high of 12-20 mGal with a width of about 20 km. It is shown that this axial gravity high is not adequately explained by uncompensated constant density topography or an isostatically compensated thermally expanded ridge crest. It can be adequately explained by Airy compensation with a depth of compensation of at least 20 km or by elastic plate compensation with a flexural rigidity between 1018 and 1020N m. The significance of the depth and flexural rigidity parameters is that in both cases they result in a gravity field from the compensation whose amplitude is below the data resolution for wavelengths less than about 50 km. That is, the gravity data do not require compensation of the rise axis for wavelengths less than about 50 km. This suggests that these wavelengths are either supported mechanically or by other forces (dynamic). The mechanical interpretation is not consistent with partial melt under the axis, indicating that dynamic forces may be important in supporting the rise axis. The same conclusion is arrived at from the thermal model. The thermal model can be made to fit the gravity data if we allow departures from isostasy and the addition of a dike-like body about 2 km wide in the crust having a positive contrast of about 0.25 Mg m-3. This geometry is consistent with a seismically determined low velocity zone in the crust in the ROSE area. In this model the nonhydrostatic pressures must be supported dynamically. A general conclusion of this study (for the portions of the East Pacific Rise studied here), is that the axial topography is not buoyantly supported by low density material in the crust. Therefore, either the crust under the rise axis mechanically supports the axial topography or a mass excess at the thermally expanded rise axis is supported by other forces, such as plate driving force. The latter option is favored on the basis of heat fluxes and seismic data.