The Swedish micro-satellite Astrid-2 contains a tri-axial fluxgate magnetometer with the sensor co-located with a Technical University of Denmark (DTU) star camera for absolute attitude, and extended about 0.9 m on a hinged boom. The magnetometer is part of the RIT EMMA electric and magnetic fields experiment built as a collaboration between the DTU, Department of Automation and the Department of Plasma Physics, The Alfvenlaboratory, Royal Institute of Technology (RIT), Stockholm. The final magnetic calibration of the Astrid-2 satellite was done at the Lovoe Magnetic Observatory under the Geological Survey of Sweden near Stockholm on the night of May 15.-16., 1997. The magnetic calibration and the intercalibration between the star camera and the magnetic sensor was performed by measuring the Earth's magnetic field and simultaneously observing the star sky with the camera. The rotation matrix between the magnetometer orthogonalized axes and the star camera optical axes was determined from the observed stellar coordinates related to the Earth magnetic field from the Magnetic Observatory. The magnetic calibration of the magnetometer integrated into the flight configured satellite was done in the (almost) constant Earth's magnetic field of about 50,000 nT by the 'Scalar Calibration Method' developed at the DTU. The satellite was positioned in 60 different attitudes relative to the Earth's magnetic field and magnetometer readings were recorded for about one minute in each position. Averages of the magnetometer readings in each position were related to the field magnitudes from the Observatory, and a least squares fit for the 9 magnetometer calibration parameters was performed (3 offsets, 3 scale values and 3 inter-axes angles). After corrections for the magnetometer digital-to-analogue converters' nonlinearities, and after fitting for a magnetic field gradient at the outdoors measurement site, the linear mathematical model of the magnetometer showed 1.26 nT rms deviation between the magnetic field measured by the observatory and the field calculated from the 60 magnetometer reading averages using the best fit calibration parameters. Owing to time shortage, we did not evaluate the temperature coefficients of the flight sensor calibration parameters. However, this was done for an identical flight spare magnetometer sensor at the magnetic coil facility belonging to the Technical University of Braunschweig over a temperature range from -47°C to +18°C. The scale values had temperature coefficients of about 11 ppm/K, axes 1 and 2 showed less than 1 nT in total offset change over the full temperature range, whereas axis 3 had 0.47 nT/K offset change. The angles between the axes changed less than 0.48 arc sec/K. The flight sensor will be assumed to have similar temperature behaviour, and the coefficients will be fitted for during the planned in-orbit magnetic calibrations.