The purpose of this study was to quantitatively evaluate biases caused by the use of ideal PRESS simulations. Metabolite basis spectra were simulated for an ideal PRESS sequence as well as with real shaped RF-pulses. Theoretical ground truth spectra were constructed for different TE and shim settings. They were fitted using both basis sets. It is shown that the fitting accuracy decreases when using ideal simulations and they depend on TE and metabolite. Therefore, simulation of basis sets should include the effects of the real pulse shapes even for the presented case of short TE and fairly large B1 amplitude.