This work compares different types of objective functionals that may be used in the formulation of the inversion methodology called Full Waveform Inversion. The behavior of the inversion scheme is related to the choice of this functional, which provides a measure that quantifies the discrepancy between the observation (real seismic data) and the prediction (simulated data corresponding to a set of model parameters) [FICHTNER, 2010]. For the inversion scheme to converge to an acceptable solution, applying gradient based methods, in a reasonable number of iterations and without being trapped in a local minimum, two factors are important: the space solution topology and its derivatives in relation to the model parameters specified. Both factors are strongly connected to the choice of the employed objective functional [FICHTNER, 2010]. The behaviors of different objective functionals were analyzed using synthetic examples based on the Marmousi model. Data sets were created considering the acoustic wave equation and presence or absence of additive random noise.