Decay heat is a key safety parameter in nuclear systems. Its accurate prediction requires reliable nuclear data and robust uncertainty quantification methods. This master thesis focuses on the propagation of fission yield uncertainties to decay heat calculations using the Monte Carlo approach, with particular attention to the effects of choosing different sampling methods, including lognormal and gamma distributions, and on the role of covariance matrices. Various sampling methods are applied using the JEFF-4.0 library, which provides official covariance matrices for thermal systems. The resulting samples are evaluated through statistical indicators, some taken from the literature and some developed specifically for this work, to assess convergence in probability and the quality of the samples produced by each method.