Two main problems are encountered in deep-tow 3C magnetic surveys. The first problem is related to instrumental error due to the manufacturing of the sensor, its integration in the towed fish structure, and the magnetization of the vehicle carrying the magnetometer; the second is related to the variation in altitude of the instruments during the dive. We evaluated a new type of calibration approach for deep-tow fluxgate magnetometers. We found that the magnetometer can be calibrated with no recourse to the vehicle attitude (pitch, roll, and heading, as it is usually achieved) but only using the three components recorded by the magnetometer and an approximation of the scalar intensity of the earth’s magnetic field. This method, called scalar calibration, allowed us to eliminate the intrinsic instrumental errors as well as the magnetization effect of the tow vehicle. Thus, despite the low maneuverability of the towed fish during the calibration experiment, we discovered a significant improvement in obtaining accurate magnetic anomaly profiles. Because only the total field anomaly and not the magnetic vector is suitable for this method, we investigated the possibility of calculating the three components via an equivalent-source approach. Therefore, assuming a 2D topographic equivalent layer, we found a stable and a meaningful magnetization of the oceanic crust. We discovered that although magnetic data are acquired along uneven tracks, this model, based on a single linear inversion, is sufficient to provide a first-order depth and magnetization intensity of the crust and also to carry out upward continuation of the total anomalous field as well as its associated vector.