The work presented in this paper is part of the broader issue of quantification of safety margins within a load-capacity framework in which uncertainties in loads and capacities are identified and quantified. The present paper describes an example of quantification of uncertainty in the capacity, i.e., the fuel failure enthalpy given a burnup level. The phenomena arising at high burnup are characterized by large uncertainties, as indicated by the scatter in the experimental data. We propose a framework for the probabilistic analysis of the failure limit, i.e., the enthalpy at failure, as a function of burnup. As an example, we obtain the distribution of the failure enthalpy for a Ziracloy-4 rod subjected to a reactivity-initiated accident in a pressurized water reactor by propagating the relevant uncertainties. We use the FRAPCON and FRAPTRAN computer codes, as well as a model for the probability of spallation, to simulate the transient and to obtain data points to derive the conditional probability distribution of the failure enthalpy at a given burnup level. The final results show that the distribution of the failure enthalpy shifts to lower values as burnup increases and that spallation is an important phenomenon.