The accuracy of a hyperthermia treatment simulation determines its role in prospective treatment planning and dosimetry for the individual patient. Of importance are (1) the accuracy of the numerical simulation method, and (2) the numerical description of the hyperthermia treatment system. The accuracy of the numerical method is generally determined by analysis of a problem that is analytically solvable. The validity of the description of the hyperthermia system is to be investigated by comparison of the simulated and measured EM-field amplitude and phase of the clinical operational hyperthermia system. In this paper, the numerical description of the AMC-4 waveguide phased array for which treatment planning is performed by the Weak Form of the Conjugate Gradient FFT method is investigated. The accuracy of the description is analysed for four set-ups, varying the bolus size and the number of waveguides attached to a phantom. The comparison of experimental and numerical data has demonstrated the ability of the Weak Formulation of the Conjugate Gradient FFT method to predict the EM-field of the AMC-4 waveguide array hyperthermia system, including effects due to bolus size variations. However, based on the comparison of the measured EM-field distributions and those obtained from simulations, the accuracy of the planning system is found to be insufficient for quantitative SAR dosimetry for individual patients. Qualitative SAR dosimetry can be applied in cases where the accuracy is of minor importance, e.g. for the retrospective analysis of problematic hyperthermia treatments. Prospective applications of the treatment planning system include the (qualitative) determination/simulation of a set of starting points giving a 'close to optimal' amplitude and phase setting, the prediction of possible problem areas and the analysis of the performance of new/improved hyperthermia devices.