Accurate simulations on detailed realistic head models are necessary to gain a better understanding of the response to transcranial magnetic stimulation (TMS). Hitherto, head models with simplified geometries and constant isotropic material properties are often used, whereas some biological tissues have anisotropic characteristics which vary naturally with frequency. Moreover, most computational methods do not take the tissue permittivity into account. Therefore, we calculate the electromagnetic behaviour due to TMS in a head model with realistic geometry and where realistic dispersive anisotropic tissue properties are incorporated, based on T1-weighted and diffusion-weighted magnetic resonance images. This paper studies the impact of tissue anisotropy, permittivity and frequency dependence, using the anisotropic independent impedance method. The results show that anisotropy yields differences up to 32% and 19% of the maximum induced currents and electric field, respectively. Neglecting the permittivity values leads to a decrease of about 72% and 24% of the maximum currents and field, respectively. Implementing the dispersive effects of biological tissues results in a difference of 6% of the maximum currents. The cerebral voxels show limited sensitivity of the induced electric field to changes in conductivity and permittivity, whereas the field varies approximately linearly with frequency. These findings illustrate the importance of including each of the above parameters in the model and confirm the need for accuracy in the applied patient-specific method, which can be used in computer-assisted TMS.