Magnetically shielded rooms for specific high resolution physiological measurements exploiting the magnetic field, e.g., of the brain (dc-magnetoencephalography), low-field NMR, or magnetic marker monitoring, need to be reproducibly demagnetized to achieve reliable measurement conditions. We propose a theoretical, experimental, and instrumental base whereupon the parameters which affect the quality of the demagnetization process are described and how they have to be handled. It is demonstrated how conventional demagnetization equipment could be improved to achieve reproducible conditions. The interrelations between the residual field and the variability at the end of the demagnetization process are explained on the basis of the physics of ferromagnetism and our theoretical predictions are evaluated experimentally.