Abstract Magnetic encephalography is currently the most informative method of functional study of the brain, since, unlike other methods, it allows one to localise deep sources of biosignals and perform three-dimensional mapping of neuronal activity. The main factors hindering the development and spread of this method are the complexity and high cost of diagnostic tools, as well as the rigidity of the requirements they impose on the spatial and temporal uniformity of the magnetic field. The prospects for desinging a device capable of largely overcoming these limitations are considered. A review of studies aimed at developing an optical sensor applicable to magnetic encephalography is presented. The all-optical single-beam nonzero-field sensor proposed by the authors earlier is separately considered.