A good knowledge of the decay heat of the various elements of the core (fissile and fertile zones, structures, sodium,..) as well as the associated uncertainties is critical for the operation of a nuclear facility, but also at the design stage as in the case of the ASTRID technological demonstrator. For this reactor, the amount of decay heat is all the more important as it will strongly impact the design of the dedicated EPur system. The uncertainties on the decay heat calculation of a sodium fast reactor that are currently used at CEA were defined several decades ago and are very high, especially at short cooling times. They may thus be penalizing for the design and operation of ASTRID. These past uncertainty assessments were based on a semi-empirical approach - a combination of physical considerations and experimental results - and on calculations using conservative methods and obsolete nuclear data libraries that need to be updated.We have used the up-to-date calculation tools in order to reduce these uncertainties ERANOS-2.2 for the neutronic calculations and DARWIN-2.3 for the depletion calculations. The uncertainty on the decay heat is estimated with the CYRUS tool, which performs the propagation of all the relevant nuclear data uncertainties (radioactive decay periods, branching ratios, cross sections, independent fission yields). The resulting total 1 sigma uncertainty on the total calculated decay heat never exceeds 2.6percent, which shows a major improvement compared to the previous uncertainty evaluation that could reach 6.75percent.