The energy calculation of the p-modes requires accurate estimates of both amplitudes and linewidths. These parameters can be derived from a suitable model used to fit the spectral distribution of the mode. In this poster we describe the technique used to achieve this goal. The model chosen to represent the power spectrum of a solar oscillation is a Lorentzian profile and, for the more general case of multiplets, a superposition of Lorentzian profiles. We use a maximum likelihood method to estimate the amplitudes and linewidths from the model and we calculate the energy from the product between the amplitude square and the linewidth. In this poster we also describe the adopted strategy to calculate the power spectrum for reducing the bias in the estimated amplitudes and linewidth. Artificial data to simulate solar oscillations are used to test the adopted strategy. We have applied this technique to three power spectra obtained from 619-day long time series provided by both MDI and GOLF experiments aboard SOHO. The amplitude, linewidth and energy are calculated for acoustic modes with l = 0 - 3, in the frequency range between 1.5 mHz and 4.0 mHz. The results are presented with particular emphasis on the differences among the investigated data sets.