In this paper, the effect of integration cap size on geoid determination was studied, when
using gravity disturbances and vertical deflection as input data. For this purpose, a series of
varying cap radii was used to predict gravimetric geoidal heights at discrete GPS-Benchmarks,
using both the Hotine and deflection-geoid techniques. In both cases, the results showed
significant dependence of the resulting geoid accuracy on the integration cap size. The two
methods showed comparable behavior in the vicinity of the cap radius, which is consistent wit? the
maximal resolution of the reference geopotential model. At larger cap sizes, the perfomance of
the deflection data type was significantly better than the gravity disturbances, which in turn
showed a dramatic degradation of the geoid accuracy. Therefore, when solving for the geoid
without modifying the integration kemel, it is strongly recommended to nse large cap sizes along
with the deflection-geoid method. If gravity disturbances are to be used for geoid determination,
then it is recommended to use as small integration cap as possible.