The reliability of the continuous transformations of origin of the current density method, which makes the transverse paramagnetic current vanish (CTOCD-PZ), for the prediction of nearly gauge-origin independent molecular magnetic susceptibility and gauge-origin independent nuclear magnetic shielding, is proved on the basis of a fairly large number of calculations. It is shown that, within the computational scheme provided by the coupled Hartree–Fock perturbation theory (CHF), convergence towards the presumed Hartree–Fock limit, for magnetic susceptibility and proton magnetic shielding, is systematically reached using basis sets which are smaller than those required by conventional common origin and CTOCD-DZ techniques. For second-row nuclear magnetic shieldings a variant of the CTOCD-PZ method, which shifts the origin of the current towards the nearest nucleus for points close to nuclei, as suggested originally by Keith and Bader with the CSDGT method [T. A. Keith and R. F. W. Bader, Chem. Phys. Lett. 210, 223 (1993)], gives likewise good results with affordable basis sets.