AbstractNew‐generation synthetic aperture radar altimetry, as implemented on CryoSat‐2, observes sea surface heights in coastal areas that were previously not monitored by conventional altimetry. Therefore, CryoSat‐2 is expected to improve the coastal mean dynamic topography (MDT). However, the MDT remains highly reliant on the geoid. Using new regional geoid models as well as CryoSat‐2 data, we determine three geodetic coastal MDT models in Norway and validate them against independent tide‐gauge observations and the operational coastal ocean model NorKyst800. The CryoSat‐2 MDTs agree on the ∼3–5 cm level with both tide‐gauge geodetic and ocean MDTs along the Norwegian coast. In addition, we compute geostrophic surface currents to help identifying errors in the geoid models. We find that even though the regional geoid models are all based on the latest satellite gravity data as provided by GOCE, the resulting circulation patterns differ. We demonstrate that some of these differences are due to erroneous or lack of marine gravity data. This suggests that there is significant MDT signal at spatial scales beyond GOCE, and that the geodetic approach to MDT determination benefits from the additional terrestrial gravity information provided by a regional geoid model. We also find that the border of the geographical mode mask of CryoSat‐2 coincides with the Norwegian Coastal Current, making it challenging to distinguish between artifacts in the CryoSat‐2 observations during mode switch and ocean signal.