Empirical mode which estimate the magnetic field direction and magnitude at any point within the magnetosphere under a variety of conditions play an important role in space weather forecasting. We report here on a number of different studies aimed at quantitatively evaluating these models, and in particular the Tsyganenko T89a model. The models are evaluated in two basic ways: (1) by comparing the range of magnetic field tilt angles observed at geosynchronous orbit with the ranges predicted for the same locations by the models; and (2) by comparing the observed magnetic field mapping between the ionosphere and geosynchronous orbit (using two-satellite magnetic field conjunctions) with the model predictions at the same locations. We find that while the T89a model predicts reasonably well the basic variation in tilt angle with local time and permits a range of field inclinations adequate to encompass the majority of observed angles on the dawn, dusk, and night sides, it is unable to reproduce the range of inclinations on the dayside. The model also predicts a smaller magnetic latitude range of geosynchronous field line footpoints than the observed two-satellite mapping indicate. Together, these results suggest that the next generation of field models should allow a greater range of stretching, especially in local time sectors away from midnight. It is important to note, however, that any increased range should encompass less-stretched configurations: although there are certainly cases where the models are not sufficiently stretched, we find that on average all magnetic field models tested, including T89a, are too stretched. Finally, in investigating how well the observed degree of field stretch was ordered by various magnetospheric indices, we find that the tilt of the field at geosynchronous orbit is a promising candidate for the incorporation into future models.