Measurement of magnetic vector or tensor quantities, namely of field or field gradient, delivers more details of the underlying geologic setting in geomagnetic prospection than a scalar measurement of a single component or of the scalar total magnetic intensity. Currently, highest measurement resolutions are achievable with SQUID-based systems. Due to technological limitations, it is necessary to suppress the parasitic magnetic field response from the SQUID gradiometer signals - which are a superposition of one tensor component and all three orthogonal magnetic field components. This in turn requires an accurate estimation of the local magnetic field. Such a measurement can itself be achieved via three additional orthogonal SQUID reference magnetometers. It is the calibration of such a SQUID reference vector magnetometer system that is the subject of this paper. A number of vector magnetometer calibration methods are described in the literature. We present two methods that we have implemented and compared, for their suitability of rapid data processing and integration into a Full Tensor Magnetic Gradiometry, SQUID-based, system. We conclude that the calibration routines must necessarily model fabrication misalignments, field offset and scale factors, and include comparison with a reference magnetic field. In order to enable fast processing on site, the software must be able to function as a stand-alone toolbox.