Implementing multicomponent diffusion models in numerical combustion stud- ies is computationally expensive due to the challenges involved in computing diffusion coefficients. As a result, mixture-averaged diffusion treatments are used to avoid these costs. However, to the authors’ knowledge, the accuracy and appropriateness of the mixture-averaged diffusion models has not been verified for three-dimensional turbulent premixed flames. This study will evaluate the role of multicomponent mass diffusion in premixed, high-Karlovitz hydrogen flames, neglecting secondary Soret and Dufour effects. Direct numerical simulation (DNS) of these flames is performed by implementing the Stefan–Maxwell equations in NGA. A semi-implicit algorithm decreases computational expense of inverting the full multicomponent ordinary diffusion array while maintaining simulation accuracy and fidelity. The algorithm is unconditionally stable pro- vided the components of the mixture-averaged and multicomponent diffusion coefficient matrices are of order one over the number of species or smaller, and performance scales approximately with the number of species squared. One-dimensional simulations of premixed hydrogen flames are performed and compared with matching cases in Cantera to verify this method. Premixed two- dimensional, unstable and three-dimensional, turbulent hydrogen flames are simulated and compared with previous mixture-averaged DNS results. Simulation conditions are carefully selected to match previously published results and ensure valid comparison. A priori analysis shows similar relative angles between the species diffusion flux vectors and the species gradient vectors between mixture-averaged and multicomponent results. Further, a posteriori analysis demonstrates negligible differences in conditional means of the fuel mass fraction and its diffusion source term against temperature for mixture-average and multicomponent transport, respectively, between the two cases.