Abstract Mathematical modelling of the relationships between the low-field magnetic anisotropy and strain are redeveloped, and the calculated data are compared with the degree of anisotropy for various rock types. It is found that the degree of anisotropy for realistic strain magnitudes is unrealistically high in the ‘passive’ model. The ‘ductile’ model, for which the degree of anisotropy depends on the magnetic grain the matrix viscosity ratio, becomes realistic at high viscosity contrasts. In the ‘line/plane’ model, the theoretical degree of anisotropy corresponds well to the natural degree of anisotropy for rocks in which the carrier of magnetic anisotropy is either magnetite or phyllosilicate minerals; by contrast, in the case of haematite and pyrrhotite, the modelled degree of anisotropy is much higher than the natural values. In the ‘viscous’ model, the calculated degree of anisotropy corresponds well to that measured in sedimentary and volcanic rocks. In the models imposing pure shear strain, the natural logarithm of the degree of anisotropy to natural strain relationship can be approximately represented by a straight line, at least for low to intermediate strains; the proportionality constant varies according to the specific model and the specific carrier of the anisotropy of magnetic susceptibility.