This paper presents a new skew angle estimation algorithm for binary document images based on the FCRM (fuzzy c-regression models) clustering method with the aim to resolve the disadvantages of low accuracy and robustness of the existing approaches. This algorithm consists of four processes. The first process transfers the input image into parallel straight lines through image analysis. The second process uses the operating window selection to accelerate the executing time. The following process magnifies the image by fast interpolation to increase the accuracy of skew angle estimation. Finally, the FCRM method is applied to estimate the skew angle. A test set of 184 documents of different kinds is used to measure the performance of the proposed algorithm. Experimental results show that the proposed method has a high precision rate for different document types; it is able to accurately estimate the skew angles that range between –89o and +89o.