We investigated the Fourier transform-based downward continuation (DWC) of airborne gravity anomalies around Taiwan assisted by topographic information. The topographic data are from the latest collections of elevations and ocean depths. The DWC employs a remove-compute-restore (RCR) procedure in which the topography is removed prior to computation and then restored to achieve stable solutions. The topographic gravity effect is evaluated point-wise using the Gaussian quadrature. A Gaussian filter with an optimal smoothing parameter reduces the noise-amplifying effect of DWC. Use of topography in DWC leads to improvements of 3 to 6 mgal of gravity on land. Surface and downward continued gravity anomalies are used to determine geoidal heights by least squares collocation (LSC) in a similar RCR procedure through the same topographic data. The accuracy of the geoidal heights at Taiwan's first-order leveling benchmarks is improved by 1 to 2 cm due to inclusion of downward continued airborne gravity.