Recently, Histograms of Oriented Gradient (HOG) are applied in face recognition. Their extension, Histograms of cooccurrence of Oriented Gradient (CoHOG), which enhance spatial information were applied in pedestrian detection problem. In this paper, CoHOG are applied on the face recognition problem. Some weighted functions for magnitude gradient are tested. We also propose a weighted approach for CoHOG, where a weight value is set for each subregion of a face image. Numerical experiments performed on Yale and ORL datasets show that (i) CoHOG has a recognition accuracy higher than HOG; (ii) using gradient magnitude in CoHOG improves recognition results; and (iii) weighted CoHOG approach improves the accuracy recognition rate. The recognition results using CoHOG are competitive with some of the state of the art methods, proving the effectiveness of CoHOG descriptor for face recognition.