Piezoelectric macro fiber composites (MFCs) are widely used for energy harvesting due to their flexibility and high electromechanical conversion efficiency. In previous research studies, numerical and analytical models have been developed for MFC-harvesters using homogenized electromechanical properties. In this paper, an analytical electroelastic coupled model, based on the mixing rules, is used to predict the output responses of an MFC-harvester coupled in 33-mode of the PZT fibers. The predicted responses exhibit variation because of the uncertainty of the parameters used in the proposed model. The research reported in this paper aims to estimate the effect of the uncertainty in the physical properties of both piezoelectric fibers and epoxy matrix on the model responses and to identify the parameters which cause large output variability. The uncertainties of the parameters are defined as the variability bounds based on the tolerances reported by the manufacturers. Global sensitivity analysis (GSA) is an appropriate method to study the effect of the uncertainty of parameters on the electromechanical response of the harvester. Two GSA are applied in this paper: (i) the Morris method based on the elementary effects measures and (ii) the variance method which consists in computing the Sobol' indices, whose goal is to identify systems parameters which have significant impact on the voltage and the power outputs of a bimorph MFC-harvester. This study leads to a qualitative comparison between the two methods which are strategically different. Both methods indicate that the elastic modulus and the density of the piezoelectric fibers, the length and the thickness of the representative volume element of the MFC-patch are the most influential parameters that affect the output voltage. It has also been shown that the order of importance of the parameters can change from the short-circuit to the open-circuit condition. Furthermore, Monte Carlo simulations are used to propagate the uncertainties of the parameters in order to determine the envelope of the variability of the responses. This study reveals the importance of considering model parameter uncertainties in the development of robust prediction tools for piezoelectric energy harvesting devices.