Nuclear safety is one of the most popular concern about nuclear energy and, for this reason, knowing the safety-related parameters, at design stage, is fundamental to reduce substantially the risk. However, especially in the nuclear field, the uncertainty sources are very widespread and, as consequence, computing with accuracy important parameters results very difficult. Here, the the effective multiplication factor keff uncertainty due to only nuclear data uncertainties is investigated. In order to perform it, different methods has been developed in the past, but the stochastic Monte Carlo method is chosen for this Thesis. The Monte Carlo method, for uncertainty propagation, has become lately attractive because of faster computation skill (and so better performance) of the new computers, since it requires usually huge computational costs.