This paper reviews the data-based approach to modeling the distant
magnetospheric magnetic field. Large amount of the spacecraft magnetometer
data, accumulated since the beginning of the space era, allowed to develop
a quantitative representation of the magnetospheric configurations, based on
measurements made within a vast region extending to the Moon's orbit. The
task of modeling the distant geomagnetic field is largely complicated both by
its dependence on seasonal/diurnal changes of the Earth's dipole tilt angle and a
significant variability caused by the solar wind fluctuations and magnetospheric
substorms. Though widely used in various experimental and theoretical studies,
the existing models provide only average and often too crude approximations
for highly variable actual field. This review discusses the fundamentals of the
next-generation models, including the problem of their calibration by the solar
wind parameters, modeling the fied from Birkeland currents, and a new approach
to fitting models against data.