In this paper, we propose a new head pose tracking method for mobile devices. Recently, head pose tracking technologies are expected to be incorporated into various situations such as robot vision. Especially, we focus on the applications using a mobile system. However, head pose estimation methods usually require a high performance hardware although the mobile systems have limited computational resource. Therefore, we propose a low cost computing algorithm of head pose tracking for the purpose of developing mobile system applications. The proposed method has two stages: (1) fast rough tracking and (2) iterative sampling stage. The proposed method has a merit of low computing cost, and is able to be applied to mobile systems. In order to confirm the effectiveness of the proposed method, we show the comparison results of two types of iterative sampling algorithms. The experimental results demonstrate that our method can be incorporated into actual mobile system applications.