Empirical magnetic field models are compared with high-altitude magnetic field measurements and results from an MHD simulation. Comparison of the T96 model and observations from GOES-8 and GOES-9 shows that if the observed solar wind and IMF parameters are used to compute the model field, the model field is more stretched than the observed field. On the other hand, if measurements made by one spacecraft are used to find the model parameters that give a best-fit field at that location, the RMS error can be reduced also at the other spacecraft four hours away in local time. Comparison of T96 model and MHD simulation results shows that the empirical models have a thinner current sheet than the MHD simulation, but that the lobe field values are quite similar to each other. Furthermore, a comparison of an event-oriented, modified T89 model and MHD simulation by Pulkkinen et al.  reveals that if the empirical model is constructed by fitting to in-situ measurements, the resulting model is very similar to the MHD simulation magnetic field. These results indicate that an efficient method of utilizing the present-day empirical models is to select model parameters based on measurements from a few individual points. (C) 2001 COSPAR. Published by Elsevier Science Ltd. All rights reserved.