In this paper, an efficient numerical model for predicting the torsional fretting wear is developed, which considers the evolution of surface profile variables with the number of fretting cycles. The major advantage of this model is that it simulates the rough surface contact problems based on a semi-analytical method (SAM). The SAM is employed for calculating the pressure distribution and the relative displacement amplitude in the contact zone, combined with updating the surface profile based on the calculated nodal wear depth using a modified Archard's equation. The discrete convolution-fast Fourier transform (DC-FFT) technique and the conjugate gradient method (CGM) are adopted to improve the solving efficiency of the contact problem. At the same time, torsional fretting wear tests are carried out under a ball-on-flat configuration to obtain the wear coefficient and wear profiles. Finally, the wear profiles predicted by the numerical model for different loading cases have been compared with the corresponding experimental results. (C) 2015 Elsevier B.V. All rights reserved.