The problem of validity of the semi-phenomenological model of collisionality dependence of the polarization current contribution into the generalized Rutherford equation for neoclassical tearing modes proposed by Wilson H R et al(1996 Plasma Phys. Control. Fusion38A149) and later discussed and analysed by Maraschek M et al(1999 Plasma Phys. Control. Fusion41L1) is considered. It is shown that the collisionality dependence of the polarization current in this model contradicts that in the theory of linear MHD instabilities. Because of this, the above model is, to some extent, ad hoc. Since this model cannot be substantiated by any theory developed up until now, it should be used with care for evaluation of the plasma behaviour in ITER. The discussed disagreement between experiments, for which interpretation of this model was used, and a recent theory of magnetic islands may serve as a stimulus for further development of the theory.