Abstract. During the last twenty years, the measurement of the Earth's electromagnetic field has been greatly improved. Old devices, such as suspended magnet variometers or the inductive sensors, are now up to date, thanks to the use of negative feedback techniques. Very sensitive instruments, like the superconducting quantum interference device (SQUID) are now available for geophysical applications. The extensive use of large arrays of magnetometers has permitted the study of the distribution of telluric flow inside the Earth's crust and the delineation of large, deep conductivity anomalies. We also have a better knowledge of the distribution of the resistivity at depth, even under the ocean bottom. But we are still very far from understanding how the telluric currents are induced in the Earth, especially at very low frequencies.R6sum~. Depuis vingt ans la mesure du champ ~lectromagn6tique nature1 s'est beaucoup am~lior6e. D'anciens instruments comme les variom~tres ~t aimant suspendu ou les capteurs inductifs ont 6t~ remis en honneur grfice ~t l'utilisation de techniques nouvelles, comme la contre-r~action de champ. D'autres appareils, extr~mement sensibles, tels que les SQUID (superconducting quantum interference device) sont maintenant disponibles pour des applications g6ophysiques. L'usage intensif de r~seaux de magn~tom~tres a permis d'6tudier la distribution des courants telluriques en profondeur et de mettre en ~vidence d'importantes anomalies de conductivit6. Nous avons aussi une meilleure connaissance de la r~partition de la r6sistivit~ ~ grande profondeur y compris sous les oc6ans. Mais nous restons tr~s ignorants des m~canismes qui r6gissent l'induction des courants telluriques ~ l'6chelle de la terre enti~re, particuli~rement aux tr~s basses fr6quences.