Compared with airborne gravimetry, a technique frequently used to infer the seafloor topography at places inaccessible to ship soundings due to the presence of ice shelf or ice mélange, airborne gravity gradiometry inherently could achieve higher spatial resolution, thus it is promising for improved inference of seafloor topography. However, its estimation capability has not been demonstrated by real projects. Theoretical analysis through admittance shows that compared with gravity disturbance, gravity gradient is more sensitive to the short-wavelength seafloor topography but diminishes faster with the increase of the distance between the seafloor and airplane, indicating its superiority is recovering short-wavelength topographic features over shallow waters. We present the first numerical experiment that estimates seafloor topography from a 0.4-km resolution, real airborne gravity gradients. It is shown that airborne gravity gradiometry can recover smaller topographic features than typical airborne gravimetry, but the estimation accuracy is only ±17 m due to the presence of subsurface density variations. The long-wavelength effect of the subsurface density variations can be removed with the aid of constraining bathymetry inside the study area, whereas the short wavelengths cannot. This study expands the applications of airborne gravity gradiometry, and helps glaciologists understand its performance in seafloor topography estimation.