Determining the direction normal to the magnetopause layer is a key step for any study of this boundary. Various techniques have been developed for this purpose. We focus here on generic residue analysis (GRA) methods, which are based on conservation laws, and the new iterative BV method, where B represents the magnetic field and V refers to the ion velocity. This method relies on a fit of the magnetic field hodogram against a modeled geometrical shape and on the way this hodogram is described in time. These two methods have different underlying model assumptions and validity ranges. We compare here magnetopause normals predicted by BV and GRA methods to better understand the sensitivity of each method on small departures from its own physical hypotheses. This comparison is carried out first on artificial data with magnetopause‐like noise. Then a statistical study is carried out using a list of 149 flank and dayside magnetopause crossings from Cluster data where the BV method is applicable, i.e., where the magnetopause involves a single‐layer current sheet, with a crudely C‐shaped magnetic hodogram. These two comparisons validate the quality of the BV method for all these cases where it is applicable. The method provides quite reliable normal directions in all these cases, even when the boundary is moving with a varying velocity, which distorts noticeably the results of most of the other methods.