Thorium-plutonium mixed oxide fuel has been of interest for a number of years as a method of plutonium disposition, in particular due to its suitability for use in light water reactors operating today. A barrier to its use is the as yet unknown effect of the larger uncertainties present in thorium nuclear data libraries when compared to more traditional fuels. Uncertainty and sensitivity analysis is performed for Th-MOX fuel in an ABWR fuel assembly in order to quantify how these higher input uncertainties affect output uncertainties in the operating parameters generated by neutronic codes. Two separate methods of uncertainty analysis are compared: a Total Monte Carlo method involving a large number of sampled data libraries to repeat the same calculation many times using W1MS, and a perturbation theory-based approach using the covariance matrices of the input nuclear data along with sensitivity coefficients generated by Serpent. The uncertainty in the Th-MOX fuel results is consistently found to be significantly greater than in either of the two reference fuels considered. The sensitivity analysis identifies the five nuclide crosssections that the effective multiplication factor is most sensitive to, and the second part of the uncertainty analysis pinpoints which of these provides the greatest contribution to this output uncertainty.