Thinning is one of the critical processes for different applications in image analysis, in particular for Optical Character Recognition (OCR) applications. The accuracy performance of OCR relies on the effectiveness of thinning algorithms. However, previously there has been little attention paid for proposing thinning algorithms for Arabic script. Also, there is a lack of quantitative performance measures of thinning techniques for Arabic script. Consequently, it is unclear which thinning algorithms are more appropriate for Arabic script. In this paper, a new thinning algorithm for Arabic script is proposed with several new performance metrics. An experiment is conducted to evaluate the proposed algorithm against two well established thinning algorithms with respect to the several proposed objective performance metrics. The experimental results show that the new algorithm has the best performance among the other two thinning algorithms.