Magnetic Resonance Spectroscopy (MRS) is an effective diagnostic technique for monitoring biochemical changes in an organism. The lineshape of MRS signals can deviate from the theoretical Lorentzian or exponential lineshape due to inhomogeneities of magnetic field applied to the patient and to patientpsilas tissue heterogeneity. We call this deviation a distortion. Using an improved damping function, we estimate the lineshape with an iterative method that finds optimal parameters. In our simulations we investigate whether estimation of the unknown distorted damping function can improve the overall result. Our method improves on that in, by including iterations, consisting of applying Hankel Singular Value decomposition (HSVD) and Nonlinear Least Squares (NLLS) to calculate the frequencies, amplitudes and phases that are used to estimate a common damping function finally applied to a metabolite database; the method is applied iteratively until the parameterspsila convergence is reached. For our simulations we used a signal with 11 undamped Lorentzian components inspired by the benchmark signal of 31P provided in. Results show that identifying the right model complexity using a convenient number of sinusoids improves significantly the iterative method. Comparison of residuals for the first and last iterations show a clear improvement. We consider the bias-variance trade-off because increasing the number of sinusoids decreases the bias but increases the variance and vice versa, so we look for an optimal order that leads to the best possible results. Although perfect estimation of the lineshape distortion is difficult, simulations are important to find a good approximation. Since parameters estimated in the first iteration are used for the next ones, a bad initial estimate will affect the final result. Iterations provided the expected successive improvements, due to the enhancement of estimations.