Over the past 70 years several deterministic mathematical models of the human middle ear have been developed for assessing its physiological process, study surgical techniques and to design hearing devices. However, experimental data have shown a large uncertain concerning the dynamic response of the middle ear over different subjects. This work aims to present a preliminary study on the middle ear uncertainties evaluating its dynamics through statistical tools. Our hypothesis for the dynamic behavior dispersion observed experimentally is based on the mechanical parameters variability among subjects. A 1D lumped parameter model of the human middle ear is combined with Monte Carlo Simulations for a probabilistic assessment. It was observed that the confidence interval of the natural frequencies rises with the mode order, where the standard deviation reaches more than 10% of the mean for the third mode. In addition, the third mode has the greater number of outliers in the boxplot analysis. Also, while the frequency response of the malleus shows the greater confidence interval above 1 [kHz], the incus frequency response presents the smaller dispersion. At last, our results suggest the large dispersion seen at low frequency in the experimental data is not caused by the mechanical parameter variation.