The aim in this paper is to perform nuclear data uncertainty decomposition in order to quantify the contribution of each of the dominant nuclear data, involved in the Special Power Excursion Reactor Test III (SPERT-III) experiments, to the overall uncertainty. This is achieved with by the SHARK-X methodology, under development at PSI, for the propagation of nuclear data uncertainties in 2D lattice calculations, using CASMO5, down to 3D core transient simulations, using SIMULATE-3K. The uncertainty decomposition was carried out for six nuclear data quantities: U-235 fission, capture and nubar; and U-238 capture, elastic and inelastic scattering. The estimated uncertainties regarding both steady-state parameters such as k-eff and static reactivity worth, as well as dynamical quantities such as total power, reactivity, fuel temperature and enthalpy are presented. Uncertainty quantification results show the dominance of U-235 nubar, U-238 and U-235 capture cross sections on the overall uncertainties on k-eff and reactivity. In addition, the uncertainties in k-eff and reactivity obtained by simultaneously perturbing the six nuclear data are practically equal to those obtained by perturbing all existing nuclear data. Concerning transient, the U-238 inelastic scattering cross section is the most dominant parameter to the overall uncertainties in the time-dependent power, reactivity, maximum nodal fuel temperature and enthalpy, at all transient stages. However, it is interesting to note that, the second dominant contribution to the overall power uncertainty in the initial excursion phase is due to the U-235 nubar, while in the power reversal phase, the U-238 capture cross-section represent the second dominant contribution, and even the first during the last moments of the transient, to the overall uncertainty.