Ultra-low-frequency (ULF) field line resonances can be used to infer the mass density along magnetospheric magnetic field lines. By specifying how mass density is distributed along the magnetic field (usually a power law as a function of distance from the Earth) and a dipole magnetic field geometry, the MHD standing wave equation can be analytically solved and mass density inferred from observed field line eigenfrequencies. However, the geometry of the Earth's magnetic field can deviate significantly from a dipole, even at relatively low L shells and on the dayside magnetosphere. This study investigates the importance of including a realistic magnetic field geometry when computing plasma mass density from observed field line eigenfrequencies. A generalized version of the toroidal mode MHD standing wave equation is solved using the Tsyganenko (2002a, 2002b) empirical magnetic field model (T01). The results are compared to those found using a dipole. We find that assuming a dipole magnetic field geometry results in an overestimation of mass density. The overestimation is larger for more disturbed levels of geomagnetic activity. Our results have important implications for the inference of heavy ions in the magnetosphere. Namely, an increase in heavy ion concentration as a result of enhanced geomagnetic activity will be exaggerated unless the proper magnetic field geometry is taken into account when calculating mass density from field line eigenfrequencies.