Short echo time in vivo STEAM 1H MR spectra (4.7 T, TE = 16 ms) of normal rat brain were fitted in the time domain using a VARPRO-like algorithm called AMARES which allows an inclusion of a large amount of prior knowledge. The prior knowledge was derived from phantom spectra of pure metabolite solutions measured under the same experimental conditions as the in vivo spectra. The prior knowledge for the in vivo spectra was constructed as follows: for each VARPRO-fitted phantom spectrum one peak (the most prominent one in the in vivo spectrum) was chosen and left unconstrained in the AMARES fitting while all the other peaks in the metabolite spectrum (i.e. their corresponding parameters--amplitudes, damping factors, frequencies and phases) were fixed to the parameter values of the unconstrained peak via amplitude and damping ratios and frequency and phase shifts. Including N-acetyl-aspartate, glutamate, total creatine, cholines, glucose and myo-inositol into the fits provided results which were in agreement with published data. An inclusion of glutamine into the set of fitted metabolites was also investigated.