The theory of the anisotropy of magnetic susceptibility assumes a linear relationship between magnetization and magnetizing field, resulting in field-independent susceptibility. This relationship is valid in diamagnetic and paramagnetic minerals by definition, and also in magnetite in the fields used in common AMS meters. Pyrrhotite, hematite and titanomagnetite may show field variation of susceptibility in the same fields and therefore in principle the linear theory is incorrect for calculating the AMS. Fortunately, the linear theory nevertheless provides accurate determination of the orientations of the principal susceptibilities and of the AMS ellipsoid shape. It gives rise to inaccurate determination of the degree of AMS. Simple techniques are suggested to overcome this problem.