This paper presents a novel approach for uncertainty propagation of neutron-induced activation cross-section measurement using unscented transformation (UT). Generally, the first-order sensitivity analysis (sandwich formula) method is used for uncertainty propagation in cross-section measurement. It is based on a linear approximation of Taylor series expansion of the function of input parameters and gives satisfactory results for smooth nonlinear functions having relatively small uncertainties. On the contrary, the UT technique is completely defined by the moments of random process and hence produces better results for error propagation in the nonlinear case with large uncertainties. The UT method is easier to implement and gives results as accurate as the sandwich formula and Monte Carlo techniques. This work examines the application of the UT method in nuclear science as an alternate to the sandwich formula and Monte Carlo methods.