GOCE gravity fields are assessed in an area around Reykjanes Ridge. Ship gravity measurements were found to be to inaccurate to determine possible improvement of GOCE gravity field models compared to the best available GRACE gravity field model. Differences between the GOCE gravity field models and EGM2008 does not appear to contain a component of the mid-ocean ridge signal. However the differences follow the Greenland coastline, which could indicate small errors in EGM 2008 there as a result of piecing together different gravity field observations. A Butterworth bandpass filter was applied to gradiometer observations at orbit height. After filtering, differences between repeat tracks with a magnitude of tens of mE are present, which can not be explained by position or attitude of the satellite. In order to reach the repeatability that can be expected from GOCE measurements, filtering methods need to improve. It was found that differences between global GRACE and GOCE gravity field models are small compared to uncertainty in crustal and upper mantle structure. Thus, geophysical inversion studies should focus on the gravity gradient observations in the instrument reference frame and at orbit height.