The purpose of this work is studying the (possible) relation between the distribution of the outliers found in the European quasi geoid data set and the magnitude of the slope (gradient) in that points, either in the same signal or in the topographic heights’ signal. As we will see, this relation is highlighted by a Montecarlo method. A particular attention has been paid to the fact that, in a so wide area, the east-west step (used to calculate the gradient magnitude in the gridded data) has to be adapted to the changing curvature of the ellipsoid.